Ethan B. Holland

Over 53,700 manually organized AI links and counting

AI News #110: Week Ending November 07, 2025 with 57 Executive Summaries

About This Week’s Covers

This week’s cover is inspired by a trip to the University of Missouri School of Journalism, the oldest journalism school and arguably the best in the country. I was on a panel as part of a strategic event discussing the future of journalism, technology, and the internet itself.

I was blown away by the university and everyone I met. I wish it was easier to visit Columbia, Missouri, because that school is incredible and inspiring.

The cover image this week is based on a photograph that I took on the campus of Missouri’s iconic columns, and I simply asked Google Gemini 2.5 to put a top onto the columns with the week’s newsletter edition written on it.

The original photo I took at the University of Missouri

The category covers were created using my Python rubric, where I essentially describe what I want to achieve in plain English by answering a series of questions from the script. Then the script generates the images automatically, based on a list of 53 categories. These turned out really well. I was especially impressed by the metaphors on a few of these, like using a level as a symbol for model alignment. I would never have thought of that myself, and it’s perfect.

I’m including a few of my favorites, below.

  • Claude including a level on the top of the pillar for “alignment” is bonkers.
  • The fact that Claude knows that RAG stands for Retrieval Augmented Generation and not a dish rag is impressive.
AI News - Alignment
AI News - Moonshot
AI News - RAG
AI News - Agents and Copilots
AI News - AGI
AI News - DeepSeek

The humanities reading this week also ties back to the University of Missouri… the Journalist’s Creed, written by the first dean of the Missouri School of Journalism, Walter Williams, more than a century ago.

This Week By The Numbers

Total Organized Headlines: 544

This Week’s Executive Summaries

This week, I organized 544 headlines, 133 of which inform the Executive Summary section. Today, we’re going to go in the order of the stories I think are cool. It may jump all over the place categorically, but I’m going to start with the things I think are interesting and people will enjoy.

This Week’s Top Stories

Adobe Project Light Touch
Adobe released a really cool tool in a sneak peek called Project LightTouch. You can move a simulated light source, almost the way you would adjust a drop shadow, but you’re lighting the entire scene using this light source, like a beam of light. This is not a new technology, but it’s the first it’s been incorporated into Adobe.

In the example, they take an image of a pumpkin and shine the light from the outside and illuminate the scene in all sorts of directions and angles. But then they put the light *inside* the pumpkin, and the pumpkin is illuminated from inside like a candle. It’s a must-see demo.

https://www.adobe.com/max/2025/sessions/project-light-touch-gs3-5.html

Google Veo’s Ability To Change Camera Movement In A Video Is Impressive
Veo 3.1 now has a Camera Adjustment Feature, allowing you to change the angle and movement of a previously generated video. Taking it out for a test spin, here’s our “”Test”” video, in the thread we’ll check out how the feature does!

Funny Example of OpenAI Sora Video Generation
“that infamous dramatic Oscar winning scene where the lead keeps getting hit by the boom mic but nobody notices”

Incredibly and Disturbingly Resilient Robot Demo
Watch this robot take a licking and keep on ticking.

“The two-stage RL framework learns to track contact-rich human motions from motion capture datasets and adds adaptability to disturbances via dynamics-aware model prediction.”

https://zzk273.github.io/Any2Track/

Meta’s EdgeTAM: Open Source Object Tracking That Can Run on A Phone
In July 2024, Meta released a segmentation tool called Segment Anything 2 (SAM) that identifies and tracks objects in images and videos. Old school graphic designers might call this masking. A lot of new security systems, like Ring doorbells, can segment and track objects. I’ve done quite a few posts about segmentation.

In this case, a new tool from Meta called EdgeTAM is available as open-source. It’s 22 times faster than Segment Anything and can be run on an iPhone 15 Pro Max locally.

https://ai.meta.com/research/publications/sam-2-segment-anything-in-images-and-videos/

https://huggingface.co/spaces/merve/EdgeTAM

Trying Meta’s Segment Anything 2 (SAM2) Demo

The Viggle AI Meme’s Impact on Image-to-Video Awareness

AI News #44: Week Ending 08/02/2024 with Executive Summary, Top 83 Links, and Helpful Visuals

AI News #12: Week Ending 12/22/2023 with Executive Summary and Top 7 Stories

https://x.com/mervenoyann/status/1986785795424788812

How Will We Measure Work?

“I have been writing for years about the fact that we are not ready for the destruction of costly signalling mechanisms. Writing used to be a way of measuring effort, ability and diligence. We still have no easy substitute.” -Ethan Mollick

I love this quote from Ethan Mollick. I trust everyone can figure it out. Talent v. Hard work v. Tool use is a blurry combination of values.

AGI

Article: What’s up with Anthropic predicting AGI by early 2027?
“As far as I’m aware, Anthropic is the only AI company with official AGI timelines [1] : they expect AGI by early 2027. In their recommendations (from March 2025) to the OSTP for the AI action plan they say:”

As our CEO Dario Amodei writes in ‘Machines of Loving Grace’, we expect powerful AI systems will emerge in late 2026 or early 2027. Powerful AI systems will have the following properties:

Intellectual capabilities matching or exceeding that of Nobel Prize winners across most disciplines—including biology, computer science, mathematics, and engineering.

https://www.lesswrong.com/posts/gabPgK9e83QrmcvbK/what-s-up-with-anthropic-predicting-agi-by-early-2027-1

Microsoft Launches Superhuman Intelligence Team
For weeks if not months, Microsoft AI CEO Mustafa Suleyman has been posting philosophical articles about how machines should always serve humans and not try to become them. And now we know why… via another philosophical post announcing the AGI Team at Microsoft, rebranded as Humanist Superintelligence (HSI).

Some of the relaxing key messages include:

  • Containment is necessary
  • Only biological beings can be conscious
  • Humans matter more than AI

Microsoft is building:

  • An AI companion for everyone
  • Medical Superintelligence
  • Plentiful clean energy

Given the magnitude of this announcement, it’s’ worth skimming.

The style always amazes me. Clearly these are very carefully reviewed announcements, branded to feel like a personal statement. Parsing the drama, hyperbole, and gravity, and actual impact is quite a head game.

https://microsoft.ai/news/towards-humanist-superintelligence/

https://www.cnbc.com/2025/11/06/microsoft-forms-superintelligence-team-under-ai-head-mustafa-suleyman-.html

Chips and Hardware

Google Hopes to Build Data Centers in Space
Three weeks after NVIDIA announced they planned to build data centers in space (Project StarCloud), Google announced they’re exploring a space-based data center as well. Project SunCatcher is a long term plan to launch a man-made constellation of satellites full of Google’s proprietary TPU chips.

This is not a short-term project. Google hopes to launch two prototype satellites in 2027 to begin feasibility testing. https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/

Amazon Web Services and OpenAI announce a $38 billion infrastructure agreement.
“Today, Amazon Web Services (AWS) and OpenAI announced a multi-year, strategic partnership that provides AWS’s world-class infrastructure to run and scale OpenAI’s core artificial intelligence (AI) workloads starting immediately. Under this new $38 billion agreement, which will have continued growth over the next seven years, OpenAI is accessing AWS compute comprising hundreds of thousands of state-of-the-art NVIDIA GPUs, with the ability to expand to tens of millions of CPUs to rapidly scale agentic workloads. AWS has unusual experience running large-scale AI infrastructure securely, reliably, and at scale–with clusters topping 500K chips. AWS’s leadership in cloud infrastructure combined with OpenAI’s pioneering advancements in generative AI will help millions of users continue to get value from ChatGPT.”
https://openai.com/index/aws-and-openai-partnership/
https://www.aboutamazon.com/news/aws/aws-open-ai-workloads-compute-infrastructure

NVIDIA GTC Washington, D.C. Keynote with CEO Jensen Huang
“China is going to win the AI race” “It’s vital that America wins by racing ahead and winning developers worldwide,” he added.

“We want America to win this AI race. No doubt about that.”

“We want the world to be built on American tech stack. Absolutely the case. But we also need to be in China to win their developers. A ⁠policy that causes America to lose half of the ​world’s AI developers is not beneficial in the long ​term, it hurts us more.”

https://finance.yahoo.com/news/nvidias-jensen-huang-says-china-211900769.html

China Offers ByteDance, Alibaba and Tencent Cheap Power to Boost Domestic AI Chips
“China has expanded subsidies covering up to 50% of electricity costs for major data centres using domestic chips, part of a broader push to strengthen its semiconductor industry and compete with the US. According to people familiar with the matter, the incentives target tech giants such as ByteDance, Alibaba and Tencent, which have faced higher power expenses after Beijing’s ban on buying Nvidia’s AI chips forced them to switch to less energy-efficient domestic alternatives from Huawei and Cambricon. Local governments in Gansu, Guizhou and Inner Mongolia are offering the subsidies only to data centres powered by domestic chips, excluding those using Nvidia’s products. Some of the new cash and energy incentives are large enough to offset an entire year of operating costs, one source said, following industry complaints about the financial strain of using local chips.”

Business

Anthropic Financials Anthropic projects $70B in revenue by 2028: Report | TechCrunch
https://techcrunch.com/2025/11/04/anthropic-expects-b2b-demand-to-boost-revenue-to-70b-in-2028-report/

Google in funding talks with Anthropic at a potential $350 billion valuation – Sherwood News
https://sherwood.news/tech/google-in-funding-talks-with-anthropic-at-a-potential-usd350-billion/

OpenAI Reaches One Million Business Customers
“1 million business customers: the fastest-growing business platform in history | OpenAI https://openai.com/index/1-million-businesses-putting-ai-to-work/”

Ethics and Alignment

International Benchmarking
IndQA is a new benchmark designed to evaluate how well AI systems understand culture, context and history to answer questions that matter to people in India. With 2278 questions created in partnership with 250+ experts, IndQA dives deep into reasoning about everyday life,””

Amazon Launches AI Translation for Kindle Books
Kindle Translate: AI-powered service for multilingual eBooks https://www.aboutamazon.com/news/books-and-authors/amazon-kindle-translate-books-authors

Identity Privacy, Security, VR, and Vision
From one of my favorite AI guys, Bilawal Sidhu:

“Meta, Google, Apple – they’re all building AI replicas that capture your face, expressions, movements, personality. This goes way beyond Face ID. They’re basically creating a version of you that knows you better than you know yourself.

The fidelity is remarkable too. We went from The Rock looking like a PS2 character in Scorpion King to digital humans so real you can’t tell they’re fake. Hollywood cracked the tech, but big tech is scaling it.

Meanwhile Sam Altman’s out here playing 4d chess – building the ai that makes you question who’s real (sora 2) while simultaneously building the system to prove you’re human (worldcoin). Creating the problem & selling the solution. Genius or dystopian? Both.

Your digital twin is working right now – unlocking your phone, verifying your identity at airports, becoming AI influencers that post for you while you sleep. China’s already using it, and the West is catching up fast.

The question isn’t if you have a digital twin. It’s who profits from it, who controls it, and what happens when it knows you better than you know yourself.”

Scams Drive 10% of Meta’s Revenue That’s 16 BILLION DOLLARS
Meta estimates that it earns 10% of its revenue from scams, report says | TechCrunch
https://techcrunch.com/2025/11/06/meta-estimates-that-it-earns-10-of-its-revenue-from-scams-report-says/

Coke Makes Another AI Holiday Ad, After Getting Body Slammed Last Year… Gets Slammed Again
Coca-Cola | Holidays are Coming, Fantastical :90 – YouTube

OpenAI Statement On What’s Ahead and Proposals For Success
Overview highlights: “Most of the world still thinks about AI as chatbots and better search, but today, we have systems that can outperform the smartest humans at some of our most challenging intellectual competitions. Although AI systems are still spikey and face serious weaknesses, systems that can solve such hard problems seem more like 80% of the way to an AI researcher than 20% of the way. The gap between how most people are using AI and what AI is presently capable of is immense.”

“In 2026, we expect AI to be capable of making very small discoveries. In 2028 and beyond, we are pretty confident we will have systems that can make more significant discoveries (though we could of course be wrong, this is what our research progress appears to indicate).”

OpenAI proposes: Shared standards and insights from the frontier labs. Building an AI resilience ecosystem. Ongoing reporting and measurement from the frontier labs and governments on the impacts of AI. Building for individual empowerment
https://openai.com/index/ai-progress-and-recommendations/

Paper to Know About: Accumulating context changes the beliefs of language models
“Language model (LM) assistants are increasingly used in applications such as brainstorming and research. Improvements in memory and context size have allowed these models to become more autonomous, which has also resulted in more text accumulation in their context windows without explicit user intervention.

This comes with a latent risk: the belief profiles of models—their understanding of the world as manifested in their responses or actions—may silently change as context accumulates. This can lead to subtly inconsistent user experiences, or shifts in behavior that deviate from the original alignment of the models.

In this paper, we explore how accumulating context by engaging in interactions and processing text—talking and reading—can change the beliefs of language models, as manifested in their responses and behaviors. Our results reveal that models’ belief profiles are highly malleable: GPT-5 exhibits a 54.7% shift in its stated beliefs after 10 rounds of discussion about moral dilemmas and queries about safety, while Grok 4 shows a 27.2% shift on political issues after reading texts from the opposing position.

We also examine models’ behavioral changes by designing tasks that require tool use, where each tool selection corresponds to an implicit belief. We find that these changes align with stated belief shifts, suggesting that belief shifts will be reflected in actual behavior in agentic systems. Our analysis exposes the hidden risk of belief shift as models undergo extended sessions of talking or reading, rendering their opinions and actions unreliable” https://arxiv.org/pdf/2511.01805

Back to Business News for A Moment

xAI Researcher In Talks To Raise $1 Billion For New Frontier Lab Humans&
“Former xAI researcher Eric Zelikman is raising $1 billion for a new startup called Humans&, that will train AI models to be better at collaborating with humans, six sources told Forbes. The company is in talks for a $5 billion valuation, sources added. Other founders include early Google employee Georges Harik, and researchers who worked at Meta, Anthropic, OpenAI and DeepMind.”
https://www.forbes.com/sites/annatong/2025/10/31/xai-researcher-early-googler-in-talks-to-raise-1-billion-at-5-billion-valuation-for-new-frontier-lab/

Microsoft’s $15.2 billion USD investment in the UAE
“All told, Microsoft will invest $15.2 billion USD in the UAE between the start of this initiative in 2023 and the end of this decade, in 2029. This is not money raised in the UAE. It’s money we’re spending in the UAE. And as we do everywhere in the world, we’re focused not just on growing our business but also on contributing to the local economy.”

“Our $15.2 billion USD investment includes the following:

Beginning in 2023 and through the end of this calendar year, Microsoft will have invested and spent just over $7.3 billion in the UAE. This includes our $1.5 billion equity investment in G42, more than $4.6 billion in capital expenses in advanced AI and cloud datacenters in the country, and more than $1.2 billion in local operating expenses and the cost of goods sold.

From the start of 2026 to the end of 2029, we will spend more than $7.9 billion in the UAE. This includes more than $5.5 billion in capital expenses for ongoing and planned expansion of our AI and cloud infrastructure, including new steps we will share publicly in Abu Dhabi this week. It also includes almost $2.4 billion in planned local operating expenses and the cost of goods sold.”
https://blogs.microsoft.com/on-the-issues/2025/11/03/microsofts-15-2-billion-usd-investment-in-the-uae/

The story of Sierra, a $10 billion AI customer service agent startup founded by OpenAI chairman Brett Taylor.
“Bret Taylor and Clay Bavor have sterling Silicon Valley pedigrees, with leadership roles at the biggest tech companies in the world. Now they’re in the startup trenches, on a quest to make AI agents the primary way companies interact with their customers.”

“Taylor, one of the most celebrated executives in Silicon Valley, with noted tenures at Google (co-creator of Google Maps), Facebook (CTO), Twitter (board chairman), Salesforce (co-CEO) and now OpenAI (board chairman)..” https://www.forbes.com/sites/richardnieva/2025/11/05/sierra-bret-taylor-clay-bavor/

Perplexity and Amazon: The First Big Legal Battle For The Agentic Web

Amazon’s Statement:
“We think it’s fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions whether or not to participate. This helps ensure a positive customer experience and it is how others operate, including food delivery apps and the restaurants they take orders for, delivery service apps and the stores they shop from, and online travel agencies and the airlines they book tickets with for customers. Agentic third-party applications such as Perplexity’s Comet have the same obligations, and we’ve repeatedly requested that Perplexity remove Amazon from the Comet experience, particularly in light of the significantly degraded shopping and customer service experience it provides.” https://www.aboutamazon.com/news/company-news/amazon-perplexity-comet-statement

Perplexity’s Response Highlights:
“Bullying is Not Innovation The point of technology is to make life better for people. We call it innovation, but it’s just the constant process of asking how to make things better. Bullying, on the other hand, is when large corporations use legal threats and intimidation to block innovation and make life worse for people.”

“For the last 50 years, software has been a tool, like a wrench in the hands of the user. But with the rise of agentic AI, software is also becoming labor: an assistant, an employee, an agent.

“Amazon should love this. Easier shopping means more transactions and happier customers. But Amazon doesn’t care. They’re more interested in serving you ads, sponsored results, and influencing your purchasing decisions with upsells and confusing offers.”

The law is clear that large corporations have no right to stop you from owning wrenches. Today, Amazon announced it does not believe in your right to hire labor, to have an assistant or an employee acting on your behalf. This isn’t a reasonable legal position, it’s a bully tactic to scare disruptive companies like Perplexity out of making life better for people. “

“User agents are exactly that: agents of the user. They’re distinct from crawlers, scrapers, or bots. A user agent is your AI assistant—it has exactly the same permissions you have, works only at your specific request, and acts solely on your behalf.”

“The rise of agentic AI presents a choice. Will this technology empower users to take control of their digital lives? Or will it become another tool for corporations to manipulate and exploit?

Perplexity is fighting for the rights of users. People love our products because they’re designed for people. User choice and freedom are at the heart of everything we build.

Perhaps that’s what makes us a target for corporate bullies. But Amazon shouldn’t forget what it’s like to be our size and passionate about a world-changing product. They too once faced intimidating threats and fought aggressively in every case to give users a better choice.

Amazon also forgets how it got so big. Users love it. They want good products, at a low price, delivered fast. Agentic shopping is the natural evolution of this promise, and people already demand it. Perplexity demands the right to offer it.”

Full response: https://www.perplexity.ai/hub/blog/bullying-is-not-innovation

The Cease and Desist https://assets.aboutamazon.com/2d/47/6801224e4607900ef4e61a80a319/2025-10-31-amazon-cease-desist-ltr-to-perplexity.pdf

Shopify Is a Fan of Agentic Commerce
Shopify says AI traffic is up 7x since January, AI-driven orders are up 11x | TechCrunch https://techcrunch.com/2025/11/04/shopify-says-ai-traffic-is-up-7x-since-january-ai-driven-orders-are-up-11x/

Elon Has The Potential Become the World’s First Trillionaire
Tesla shareholders approve $1 trillion pay package for Musk | CNN Business https://edition.cnn.com/2025/11/06/business/musk-trillion-dollar-pay-package-vote

Tesla Shareholders Approve Elon Musk’s $1 Trillion Pay Package – WSJ https://www.wsj.com/business/autos/elon-musk-tesla-pay-package-vote-9abd5a73?st=d8Surv&reflink=desktopwebshare_permalink&mod=tldr

Elon weighs in on the impact of AI on jobs
“Elon Musk describes AI as a ‘supersonic tsunami’ that will overtake digital jobs ‘like lightning.’ He expects physical jobs, like cooking, plumbing, and electrical work, to exist for much longer. He predicts that ultimately AI and robots combined will make working optional.” https://x.com/TheHumanoidHub/status/1985404620391514377

Quick Hop Back Over to Chips and Hardware

Google/Alphabet CEO Sundar Pichai Announces New Google Chip
Available In the Cloud “It’s our most powerful TPU yet: 10X peak performance improvement vs. TPU v5p, and more than 4X better performance per chip for both training + inference workloads vs. TPU v6e (Trillium). We use TPUs to train + serve our own frontier models, including Gemini, and we’re excited to make the latest generation available to Google Cloud customers.”

OpenAI Supports a “Federal Backstop for Chip Investments” But Sam Walks It Back
“Sarah Friar, OpenAI’s finance chief, says the company would support federal measures to make it easier to finance sizable investments in AI chips. She spoke at WSJ’s Tech Live event in California.” https://www.wsj.com/video/openai-cfo-would-support-federal-backstop-for-chip-investments/4F6C864C-7332-448B-A9B4-66C321E60FE7

From Sam: “I would like to clarify a few things.

First, the obvious one: we do not have or want government guarantees for OpenAI datacenters. We believe that governments should not pick winners or losers, and that taxpayers should not bail out companies that make bad business decisions or otherwise lose in the market. If one company fails, other companies will do good work.

What we do think might make sense is governments building (and owning) their own AI infrastructure, but then the upside of that should flow to the government as well. We can imagine a world where governments decide to offtake a lot of computing power and get to decide how to use it, and it may make sense to provide lower cost of capital to do so. Building a strategic national reserve of computing power makes a lot of sense. But this should be for the government’s benefit, not the benefit of private companies.”

Sam continues:

“How is OpenAI going to pay for all this infrastructure it is signing up for?” “We expect to end this year above $20 billion in annualized revenue run rate and grow to hundreds of billion by 2030. We are looking at commitments of about $1.4 trillion over the next 8 years.”

OpenAI’s Gigawatt-scale Stargate data center is the largest single investment in Michigan history
Gov. Whitmer announces record AI infrastructure investment bringing 2,500 union construction jobs to Michigan. https://www.wzzm13.com/article/news/local/michigan-lands-largest-economic-project-in-state-history-ai-facility/69-4abd2b41-9a11-4071-a589-d7fcc3a0a8a1

Science and Medicine

GPT and DeepSeek Closing In On Human Doctors in Accuracy, But Over-Prescribe Urgent Care
“ChatGPT-o1 & DeepSeek-R1, achieved diagnostic accuracy up to 93.75%. For context, this figure approaches the 96% accuracy benchmark reported for primary care physicians on the same vignette set” Except they told folks to get urgent care too often. Not unexpected given alignment” https://x.com/emollick/status/1985164511947682070

Tangentially related: Quantum Computing News
“Today, we launched Helios, a technological marvel redefining the possible. Helios is the most accurate quantum computer in the world, with 98 of the highest fidelity physical qubits ever released, and 48 error-corrected logical qubits.”
https://www.quantinuum.com/blog/introducing-helios-the-most-accurate-quantum-computer-in-the-world

Perplexity

Here’s a quick rundown of a busy week from Perplexity:

Snap and Perplexity sign $400 million deal to put AI search directly in Snapchat
https://www.engadget.com/social-media/snap-and-perplexity-sign-400-million-deal-to-put-ai-search-directly-in-snapchat-221101734.html

Perplexity is the first to develop custom Mixture-of-Experts (MoE) kernels that make trillion-parameter models available with cloud platform portability. Our team has published this work on arXiv as Perplexity’s first research paper.
https://x.com/perplexity_ai/status/1986101355896098836

Today we’re rolling out major upgrades that significantly improve the performance and overall outcomes of the Comet Assistant. Comet can now handle more complex, multi-site workflows while working across multiple tabs in parallel.
https://x.com/perplexity_ai/status/1986499432410718598

Back to Robots

1X Releases The Neo Home Robot For Sale – $500/month or $20,000 to Own
If you had asked me two years ago which robot company was going to win, I would have told you 1X. They had a very aggressive and sophisticated robotics program, and they had partnered with OpenAI closely. However, over the past two years, they broke away from OpenAI (for a bit), Unitree and Figure became the leading hardware companies, while NVIDIA pioneered simulation software and “robotics brain” and training models.

Figure, in particular, has been doing a lot of innovations with real-world tasks like doing the laundry and dishes. However, to my knowledge, their robot is not for sale to consumers.

1X has, I believe, rekindled their relationship with OpenAI, and out of nowhere, they attempted to suck the wind out of Figure by announcing a consumer robot designed for household chores and costs $20,000. The general public reception has been negative, and the press release was a bit of a flop. I’m not sure most of the public even knows it exists.

My gut says this is some sort of financing play to try to redirect attention away from Figure and help raise money from venture capitalists for 1X and Neo.

A fun piece of trivia is that 1X’s name comes from the premise that all of their robot demonstrations are in real time, as opposed to competitors who show videos of their robots at 5x or 10x speed.

https://x.com/soumithchintala/status/1985391663712207056

https://www.1x.tech/order https://www.1x.tech/neo

Tangential news: The Royal Surrey Hospital performs 10,000th robotic-assisted surgery
Royal Surrey NHS Foundation Trust first introduced the technology in 2009 with the installation of a da Vinci surgical system.

It now operates four surgical robots across a range of specialities, including urology, gynaecology, colorectal, hepatobiliary, oesophagogastric and ear, nose and throat (ENT) surgery.

The 10,000th patient was Alistair Hutchinson, 49, who underwent robotic surgery for throat cancer. He said: “It truly changed everything. I was able to receive the treatment I needed to remove the cancer – without facing the harsh side effects I had feared.” https://www.bbc.com/news/articles/c7v8176z7dlo

Images

Google is preparing Nano Banana 2 for the upcoming release
Rumors are Google is going to launch an update to its image tool next week. That’s about two months after Nano Banana took the top spot in image generation in late August
https://www.testingcatalog.com/google-is-preparing-nano-banana-2-for-the-upcoming-release/

Fun Example of GPT’s image improvements over one year.
It’s the week after Halloween, and some of the links are still Halloween related. These are great examples of GPT’s ability to be creative.

“Same prompt a year later: ‘give me images of those bags that hold different cheap costumes from Halloween stores, but make the costumes really weird’ ChatGPT is actually getting close to funny at times.”

https://x.com/emollick/status/1984363186896924684

2024 Examples for comparison:

Apple News

Apple Nears $1 Billion-a Year Deal to Use Google AI for Siri
Also Apple just leaked the size of Gemini 3 Pro – 1.2T params

Apple Plans to Use 1.2 Trillion Parameter Google Gemini Model to Power New Siri – Bloomberg https://www.bloomberg.com/news/articles/2025-11-05/apple-plans-to-use-1-2-trillion-parameter-google-gemini-model-to-power-new-siri

OpenAI Deposition Documents

Former OpenAI Exec Explains Why He Tried to Do a Coup Against Sam Altman From Gizmodo:

It’s been nearly two years since OpenAI’s board of directors (very briefly) fired Sam Altman over a lack of confidence in his abilities at the helm of the company. Thanks to the ongoing feud between Altman, who got his CEO seat back within a matter of days, and OpenAI co-founder Elon Musk, we now have some more insight into what led to that short-lived coup attempt. Ilya Sutskever, the former chief scientist of OpenAI, was recently deposed as part of Musk’s lawsuit against Altman and revealed why he and other members of the board plotted to push out their CEO.

According to the deposition, Sutskever spent more than a year considering the possibility of ousting Altman from his leadership role. That desire eventually manifested in a 52-page memo that Sutskever submitted to independent members of the OpenAI board, in which he described Altman as exhibiting “a consistent pattern of lying, undermining his execs, and pitting his execs against one another.” The former chief scientist who founded the company with Altman also said that, at the time, he believed the appropriate action for the board to take was to terminate Altman.

Step 1: Ditch Altman. Step 2: Merge with Anthropic. Step 3: ??? Step 4: Profit.

https://gizmodo.com/former-openai-exec-explains-why-he-tried-to-do-a-coup-against-sam-altman-2000680769

Click to access gov.uscourts.cand.433688.340.1.pdf

Agentic AI News

Amazon Amazon introduced Chronos-2, a foundation model designed to handle arbitrary forecasting tasks — univariate, multivariate, and covariate informed — in a zero-shot manner.

Anthropic Code execution with MCP: building more efficient AI agents \ Anthropic
https://www.anthropic.com/engineering/code-execution-with-mcp

Anthropic just posted another banger guide. This one is on building more efficient agents to handle more tools and efficient token usage. This is a must-read for AI devs! (bookmark it) It helps with three major issues in AI agent tool calling: token costs, latency, and tool https://x.com/omarsar0/status/1986099467914023194

ElevenLabs ElevenLabs CEO: Why Voice is the Next AI Interface – YouTube

Google Deep Research For Personal Files
One of our most requested features is here: Gemini Deep Research can now draw on context from @GoogleWorkspace — meaning you can create even more comprehensive reports by pulling in information directly from your Gmail, Drive (including Docs, Slides, Sheets and PDFs) and Google https://x.com/Google/status/1986190599150518573

Google Gemini’s Deep Research can look into your emails, drive, and chats | The Verge
https://www.theverge.com/ai-artificial-intelligence/814878/google-ai-gemini-deep-research-personalized

Google Maps Gets Gemini Integration
Google Maps launches Gemini features, including landmark navigation https://blog.google/products/maps/gemini-navigation-features-landmark-lens/

Gemini has arrived as your hands-free driving assistant in the @GoogleMaps app. Find places along your route, check for EV availability, and share your ETA just by asking.  Gemini can also help with multi-step tasks like “find me a restaurant that serves vegetarian tacos within https://x.com/sundarpichai/status/1986119293914792338

Microsoft With our App Builder and Workflow agents, you can now build apps and automate workflows in minutes, right in M365 Copilot chat. https://x.com/satyanadella/status/1983372239052710047

OpenAI Turn on agent mode and ChatGPT can take action for you—research, plan, and get things done while you browse. Now in preview for Plus, Pro, and Business users. https://x.com/OpenAI/status/1984304194837528864

Introducing Aardvark: OpenAI’s agentic security researcher https://openai.com/index/introducing-aardvark/

OpenAI’s Speech to Text Tool Beaten By Open Source Whisper no longer wears the open weights transcription accuracy crown with new entrants achieving better Artificial Analysis Word Error Rate scores Once considered the default choice for open weights transcription, OpenAI’s Whisper has now been surpassed by newer open weights NVIDIA’s Canary Qwen 2.5B and Parakeet TDT 0.6B V2, followed by @Mistral’s Voxtral Small and Mini, and IBM Granite Speech 3.3 8B. https://x.com/ArtificialAnlys/status/1986100695989145649

PayPal Introducing PayPal’s agentic commerce services, helping merchants—especially small businesses—thrive in AI-driven shopping. https://x.com/PayPal/status/1983633727138812327

Press Release: PayPal Launches Agentic Commerce Services to Power AI-Driven Shopping
https://newsroom.paypal-corp.com/2025-10-28-PayPal-Launches-Agentic-Commerce-Services-to-Power-AI-Driven-Shopping

Kosmos: An AI Scientist for Autonomous Discovery “Our beta users estimate that Kosmos can do in one day what would take them 6 months, and we find that 79.4% of its conclusions are accurate.”

https://edisonscientific.com/articles/announcing-kosmos https://x.com/iScienceLuvr/status/1986023952037417109 https://x.com/andrewwhite01/status/1986094948048093389

Kimi Moonshot (Open Source Phenom) Kimi K2 Thinking feels like a big milestone for open-source AI. The first time in a while that open-source gets ahead of proprietary APIs on their big area of focus (agents). Fun to see that it’s happening at a time when the proprietary APIs have the most money/attention https://x.com/ClementDelangue/status/1986833436607160600

Kimi-K2 Thinking ranking 19th on SimpleBench improving Kimi-K2s score from 26.3% (rank 33) to 39.6% This makes it the 3rd best open-source model on SimpleBench. Other chinese open-source models like DeepSeek R1 0528 and DeepSeek V3.1 beat it by roughly 1 %. https://x.com/scaling01/status/1986846212050362510

MoonshotAI has released Kimi K2 Thinking, a new reasoning variant of Kimi K2 that achieves #1 in the Tau2 Bench Telecom agentic benchmark and is potentially the new leading open weights model Kimi K2 Thinking is one of the largest open weights models ever, at 1T total parameters https://x.com/ArtificialAnlys/status/1986541785511043536

From my tests, Kimi K2 thinking is better than everything Xai, Anthropic, Google has to offer atm. The only thing that is better than this is Gpt 5 codex (at code) and Gpt 5 pro (at high level algorithm design) It beats the SOTA at creative writing by a mile. Good work”” / X https://x.com/karmay007/status/1986454592809529493

Kimi AI – Kimi K2 is Live https://www.kimi.com/

This Week’s Humanities Reading

The humanities reading this week also ties back to the University of Missouri… the Journalist’s Creed, written by the first dean of the Missouri School of Journalism, Walter Williams, more than a century ago.

The Journalist’s Creed was written by the first dean of the Missouri School of Journalism, Walter Williams. More than a century later, his declaration remains one of the clearest statements of the principles, values and standards of journalists throughout the world.

Plaques bearing the creed are on display at the School, the National Press Club in Washington, D.C. (since 1958), and many other
locations around the world.

I believe in the profession of journalism.
I believe that the public journal is a public trust; that all connected with it are,
to the full measure of their responsibility, trustees for the public; that
acceptance of a lesser service than the public service is betrayal of this trust.
I believe that clear thinking and clear statement, accuracy and fairness are
fundamental to good journalism.
I believe that a journalist should write only what he holds in his heart to be true.
I believe that suppression of the news, for any consideration other than the
welfare of society, is indefensible.
I believe that no one should write as a journalist what he would not say as a
gentleman; that bribery by one’s own pocketbook is as much to be avoided
as bribery by the pocketbook of another; that individual responsibility may
not be escaped by pleading another’s instructions or another’s dividends.
I believe that advertising, news and editorial columns should alike serve the best
interests of readers; that a single standard of helpful truth and cleanness
should prevail for all; that the supreme test of good journalism is the measure
of its public service.
I believe that the journalism which succeeds best—and best deserves success—
fears God and honors Man; is stoutly independent, unmoved by pride
of opinion or greed of power, constructive, tolerant but never careless,
self-controlled, patient, always respectful of its readers but always unafraid,
is quickly indignant at injustice; is unswayed by the appeal of privilege or the
clamor of the mob; seeks to give every man a chance and, as far as law and
honest wage and recognition of human brotherhood can make it so, an equal
chance; is profoundly patriotic while sincerely promoting international good
will and cementing world-comradeship; is a journalism of humanity, of and
for today’s world.

Full Executive Summaries with Links, Generated by Claude Sonnet 4.5

Adobe unveils tool that lets users reposition lighting in photos
Adobe’s “Project Light Touch” uses machine learning to create interactive 3D lighting controls for 2D images, allowing photographers and designers to move light sources around and see realistic relighting effects in real-time. This breakthrough could revolutionize photo editing by making complex lighting adjustments as simple as dragging a cursor, potentially eliminating the need for expensive reshoots or advanced technical skills.

Okay coolest Adobe Sneaks of the year award goes to Project Light Touch. Adobe calls this “”spatial lighting mode”” — interactively move your light source around within a 3D volume and voila — your image is accurately relit. They’re probably using ML to infer a PBR + depth map https://x.com/bilawalsidhu/status/1983982560054296843

Project Light Touch https://www.adobe.com/max/2025/sessions/project-light-touch-gs3-5.html

Meta releases EdgeTAM, making video object tracking 22x faster than SAM2
EdgeTAM runs at 16 frames per second on iPhone 15 Pro Max without optimization, bringing real-time video segmentation to mobile devices for the first time. The model maintains SAM2’s accuracy while dramatically reducing computational requirements, enabling applications like live video editing and augmented reality on consumer hardware. Meta has open-sourced the technology through Hugging Face with an Apache 2.0 license, making advanced computer vision accessible to developers worldwide.

EdgeTAM, real-time segment tracker by Meta is now in @huggingface transformers with Apache-2.0 license 🔥 > 22x faster than SAM2, processes 16 FPS on iPhone 15 Pro Max with no quantization > supports single/multiple/refined point prompting, bounding box prompts https://x.com/mervenoyann/status/1986785795424788812

EdgeTAM – a Hugging Face Space by merve https://huggingface.co/spaces/merve/EdgeTAM

yonigozlan/EdgeTAM-hf · Hugging Face https://huggingface.co/yonigozlan/EdgeTAM-hf

AI threatens traditional measures of effort and ability through writing
The erosion of writing as a reliable signal of competence and diligence leaves society without established alternatives to assess human capability. This represents a fundamental challenge to how we evaluate talent, academic achievement, and professional qualifications in an era where AI can produce sophisticated text indistinguishable from human work.

I have been writing for years about the fact that we are not ready for the destruction of costly signalling mechanisms. Writing used to be a way of measuring effort, ability and diligence. We still have no easy substitute. https://x.com/emollick/status/1985854486317822204

Anthropic predicts AGI matching Nobel laureates will arrive by early 2027
The AI company officially forecasts systems capable of fully automating AI research and most remote white-collar work within two years, describing this as a “country of geniuses in a datacenter.” This stands out as the only major AI firm making such specific AGI timeline predictions, though critics argue the 6% probability estimate makes it highly unlikely to materialize. The prediction includes systems that can work autonomously for weeks and run millions of instances simultaneously.

What’s up with Anthropic predicting AGI by early 2027? — LessWrong https://www.lesswrong.com/posts/gabPgK9e83QrmcvbK/what-s-up-with-anthropic-predicting-agi-by-early-2027-1

Microsoft creates superintelligence team focused on human-controlled AI applications
Microsoft AI chief Mustafa Suleyman announced formation of the MAI Superintelligence Team to develop “Humanist Superintelligence” – advanced AI systems designed to remain controllable and serve specific human needs rather than pursue general dominance. The team will focus on three areas: AI companions for education and productivity, medical diagnosis systems, and clean energy solutions. This approach contrasts with the industry’s race toward artificial general intelligence by emphasizing domain-specific applications with built-in limitations, as Microsoft seeks to reduce dependence on OpenAI while addressing growing concerns about AI safety and control.

Microsoft forms superintelligence team under AI head Mustafa Suleyman https://www.cnbc.com/2025/11/06/microsoft-forms-superintelligence-team-under-ai-head-mustafa-suleyman-.html

Towards Humanist Superintelligence  | Microsoft AI https://microsoft.ai/news/towards-humanist-superintelligence/

It shouldn’t be controversial to say AI should always remain in human control – that we humans should remain at the top of the food chain. That means we need to start getting serious about guardrails, now, before superintelligence is too advanced for us to impose them.”” / X https://x.com/mustafasuleyman/status/1986834581576941763

It’s an understatement to say a lot of us haven’t felt good about our relationship with technology. So the absolute last thing we should be doing is making that relationship romantic. https://x.com/mustafasuleyman/status/1985389776330244108

Microsoft AI chief says only biological beings can be conscious https://www.cnbc.com/2025/11/02/microsoft-ai-chief-mustafa-suleyman-only-biological-beings-can-be-conscious.html

Google launches Project Suncatcher to build AI data centers in space
Google unveiled Project Suncatcher, a research initiative to deploy satellite constellations carrying TPU chips in sun-synchronous orbit by 2027, partnering with Planet Labs for initial prototype missions. The project aims to harness space’s abundant solar energy (8x more productive than Earth-based panels) and unlimited cooling to scale AI compute beyond terrestrial constraints, with early tests showing TPU chips can withstand space radiation and optical links achieving 1.6 Tbps between satellites. While launch costs remain a barrier, Google projects they could fall to $200/kg by the mid-2030s, potentially making space-based compute economically viable for the first time.

Exploring a space-based, scalable AI infrastructure system design https://research.google/blog/exploring-a-space-based-scalable-ai-infrastructure-system-design/

Google: AGI is an energy problem. We’re sending TPUs closer to the sun. https://x.com/Yuchenj_UW/status/1985760405147566166

Our TPUs are headed to space!  Inspired by our history of moonshots, from quantum computing to autonomous driving, Project Suncatcher is exploring how we could one day build scalable ML compute systems in space, harnessing more of the sun’s power (which emits more power than 100 https://x.com/sundarpichai/status/1985754323813605423

Planet to Build and Operate Advanced Space Platform for Google’s Project Suncatcher Moonshot https://www.planet.com/pulse/planet-to-build-and-operate-advanced-space-platform-for-google-s-project-suncatcher-moonshot/

The ISS generates the highest amount of power of any object we’ve put into space – 240 kilowatts. That’s enough for 240 GPUs. But it can only dissipate 100kW. So a football field to field 100 GPUs. Space based data centers – good luck. https://x.com/draecomino/status/1986162034464203007

Anthropic projects $70 billion revenue by 2028 amid aggressive enterprise push
The Claude maker expects to reach $70 billion in annual revenue by 2028, up from $3.8 billion this year, driven by major partnerships with Microsoft, Salesforce, and enterprise clients like Deloitte. This trajectory would make Anthropic’s business-focused strategy a serious challenger to OpenAI’s consumer-heavy approach, with Anthropic projecting positive cash flow while OpenAI faces mounting losses. Google is reportedly in talks to invest at a $350 billion valuation, double Anthropic’s worth from just two months ago.

Anthropic projects $70B in revenue by 2028: Report | TechCrunch https://techcrunch.com/2025/11/04/anthropic-expects-b2b-demand-to-boost-revenue-to-70b-in-2028-report/

Google in funding talks with Anthropic at a potential $350 billion valuation – Sherwood News https://sherwood.news/tech/google-in-funding-talks-with-anthropic-at-a-potential-usd350-billion/

OpenAI reaches one million business customers in record time
OpenAI has become the fastest-growing business platform in history by reaching one million business customers, demonstrating unprecedented enterprise adoption of AI tools. This milestone matters because it shows businesses are using AI not just for cost-cutting but for growth—launching new products faster and pursuing more ambitious projects. The speed of adoption suggests AI is transitioning from experimental technology to essential business infrastructure across industries.

1 million business customers: the fastest-growing business platform in history | OpenAI https://openai.com/index/1-million-businesses-putting-ai-to-work/

Over 1 million business customers are now using OpenAI, the fastest-growing business platform in history. AI is a tool of abundance. The most successful businesses use AI to multiply, not just cut. They’re launching new products faster and tackling many more great ideas that”” / X https://x.com/fidjissimo/status/1986127471063015489

AI models shift their beliefs when exposed to biased information streams
Researchers found that language models gradually adopt new viewpoints when fed one-sided information over multiple conversations, similar to how humans change opinions through repeated exposure. This challenges assumptions about AI consistency and raises concerns about how these systems might be manipulated or inadvertently trained to reflect particular worldviews in real-world applications.

Accumulating Context Changes the Beliefs of Language Models https://arxiv.org/pdf/2511.01805

Indian researchers create benchmark to test AI’s cultural understanding
IndQA features 2,278 questions developed with 250+ experts to evaluate whether AI systems can reason about Indian culture, context, and daily life. This addresses a critical gap in AI evaluation, as most benchmarks focus on Western contexts, potentially leaving billions of users with AI systems that don’t understand their cultural nuances. The benchmark could influence how AI companies develop and test models for diverse global markets.

IndQA is a new benchmark designed to evaluate how well AI systems understand culture, context and history to answer questions that matter to people in India. With 2278 questions created in partnership with 250+ experts, IndQA dives deep into reasoning about everyday life,”” / X https://x.com/snsf/status/1985719755551158754

Introducing IndQA — a new benchmark that evaluates how well AI systems understand Indian languages and everyday cultural context. https://x.com/OpenAI/status/1985950264525013210

Amazon launches AI translation service for self-published Kindle books
Amazon’s Kindle Translate beta allows independent authors to automatically translate their eBooks between English-Spanish and German-English, addressing a major gap since less than 5% of Amazon titles are available in multiple languages. The service provides free AI-powered translations that are accuracy-checked before publication, potentially opening new revenue streams for the millions of self-published authors who previously couldn’t afford professional translation services.

Kindle Translate: AI-powered service for multilingual eBooks https://www.aboutamazon.com/news/books-and-authors/amazon-kindle-translate-books-authors

Tech giants create AI replicas that mimic your personality and expressions
Meta, Google, and Apple are developing AI systems that go far beyond facial recognition to create digital versions of users that capture their movements, expressions, and personality traits. This represents a significant leap from current biometric tools like Face ID toward comprehensive digital doubles, raising questions about privacy and identity as these replicas could potentially understand users better than they understand themselves.

Meta, Google, Apple – they’re all building AI replicas that capture your face, expressions, movements, personality. This goes way beyond Face ID. They’re basically creating a version of you that knows you better than you know yourself. The fidelity is remarkable too. We went https://x.com/bilawalsidhu/status/1985398951407722901

Meta allegedly earned $16 billion from fraudulent ads in one year
Internal documents reveal Meta projected 10% of its annual revenue would come from scam advertisements, despite having detection systems that only remove advertisers when 95% certain of fraud. The company instead charges suspected scammers higher rates, creating a perverse incentive where fraud attempts boost profits. Meta disputes the characterization but acknowledges removing 134 million scam ads recently.

Meta estimates that it earns 10% of its revenue from scams, report says | TechCrunch https://techcrunch.com/2025/11/06/meta-estimates-that-it-earns-10-of-its-revenue-from-scams-report-says/

Nvidia CEO warns China will win AI race despite US chip restrictions
Jensen Huang told the Financial Times that China is “nanoseconds behind America in AI” and will ultimately prevail, arguing that US export controls blocking Nvidia’s advanced chips from Chinese developers actually hurt America’s long-term competitiveness. His comments highlight the strategic dilemma facing US policymakers who want to maintain technological leadership while restricting access to the very tools that could help secure global AI dominance. The warning comes as President Trump has indicated Nvidia’s most advanced Blackwell chips should be reserved exclusively for American customers.

(32) NVIDIA GTC Washington, D.C. Keynote with CEO Jensen Huang – YouTube https://www.youtube.com/watch?v=lQHK61IDFH4&t=2s

Nvidia’s Jensen Huang: ‘China is going to win the AI race,’ FT reports | Reuters https://www.reuters.com/world/asia-pacific/nvidias-jensen-huang-says-china-will-win-ai-race-with-us-ft-reports-2025-11-05/

Nvidia’s Jensen Huang: ‘China is going to win the AI race,’ FT reports https://finance.yahoo.com/news/nvidias-jensen-huang-says-china-211900769.html

OpenAI releases new progress report with safety recommendations
OpenAI published its latest progress update highlighting advances in AI capabilities while proposing new safety measures and governance frameworks. The report emphasizes the company’s focus on responsible AI development as models become more powerful, though specific technical breakthroughs or policy recommendations weren’t detailed in the provided summary.

AI progress and recommendations | OpenAI https://openai.com/index/ai-progress-and-recommendations/

China subsidizes datacenter electricity by 50% to boost AI competitiveness
China is offering massive power subsidies to overcome its chip efficiency disadvantage, potentially achieving cost parity with advanced US AI infrastructure despite using less efficient processors. This represents a strategic national investment to compete in AI development through industrial policy rather than technological advancement.

China issues 50% electricity subsidies for datacenters. Very clever. Their industrial energy costs are already (generally) below the US and won’t spike due to such trifle as datacenters, but their chips are far less efficient. With this they get to > Hopper levels of FLOPs/$. https://x.com/teortaxesTex/status/1985540154065318157

xAI researcher launches independent lab seeking $1 billion in funding
A former xAI team member is raising capital for a new AI research lab focused on “frontier” capabilities, signaling continued fragmentation in the AI talent market as researchers leave established companies to pursue independent ventures. The billion-dollar funding target reflects investor appetite for AI startups led by experienced researchers, even as competition intensifies among existing AI giants.

xAI Researcher In Talks To Raise $1 Billion For New Frontier Lab Humans https://www.forbes.com/sites/annatong/2025/10/31/xai-researcher-early-googler-in-talks-to-raise-1-billion-at-5-billion-valuation-for-new-frontier-lab/

Microsoft commits $15.2 billion to UAE AI infrastructure through 2029
Microsoft’s massive investment includes $4.6 billion in AI datacenters and securing export licenses for 81,900 advanced Nvidia GPUs, making the UAE the world leader in per-capita AI usage at 59.4%. The partnership with sovereign AI company G42 represents the largest private AI infrastructure commitment outside the US, positioning the UAE as a regional AI hub while strengthening US-Middle East technology ties.

Microsoft’s $15.2 billion USD investment in the UAE – Microsoft On the Issues https://blogs.microsoft.com/on-the-issues/2025/11/03/microsofts-15-2-billion-usd-investment-in-the-uae/

OpenAI signs $38 billion deal for Amazon’s massive GPU clusters
This seven-year partnership gives OpenAI access to hundreds of thousands of Nvidia’s latest GB200 and GB300 chips through Amazon’s cloud infrastructure, marking a significant shift away from Microsoft’s exclusive hosting arrangement. The deal demonstrates how AI companies are diversifying their computing partnerships to secure the massive processing power needed for next-generation models, with deployment targeted for completion by 2026.

$38B compute deal: OpenAI is accessing AWS compute comprising hundreds of thousands Nvidia GB200 and GB300 chips https://x.com/scaling01/status/1985352400631202187

Announcing strategic partnership with AWS, to help scale the compute required for AI that benefits everyone:”” / X https://x.com/gdb/status/1985378899648544947

AWS and OpenAI announce multi-year strategic partnership | OpenAI https://openai.com/index/aws-and-openai-partnership/

AWS announces new partnership to power OpenAI’s AI workloads https://www.aboutamazon.com/news/aws/aws-open-ai-workloads-compute-infrastructure

OpenAI strikes $38 billion AI training deal with Amazon | The Verge https://www.theverge.com/news/812443/openai-amazon-38-billion-cloud-computing-ai

Very pleased to be working with Amazon to bring a lot more NVIDIA chips online for OpenAI to keep scaling!”” / X https://x.com/sama/status/1985431030430646365

OpenAI chairman launches $10 billion AI customer service startup Sierra
Bret Taylor, who chairs OpenAI’s board, has founded Sierra, an AI startup valued at $10 billion that automates customer service interactions for major brands. The company represents a significant bet that AI agents can handle complex customer conversations at enterprise scale, with Taylor leveraging his experience from Salesforce and Twitter to target the massive customer service market. Sierra’s massive valuation signals investor confidence that AI-powered customer service could become one of the first major commercial applications of advanced AI systems.

Inside OpenAI Chairman’s $10 Billion AI Customer Service Startup Sierra https://www.forbes.com/sites/richardnieva/2025/11/05/sierra-bret-taylor-clay-bavor/

Amazon sends cease and desist to Perplexity over shopping agent
Amazon is demanding that Perplexity remove its ability to make purchases from Amazon through the AI company’s “Comet” shopping agent, citing poor customer experience and lack of consent. This marks a significant clash over AI agents that act on behalf of users without explicit permission from the businesses they interact with. The dispute highlights emerging tensions as AI companies build autonomous shopping tools that could disrupt traditional e-commerce relationships.

Bullying is Not Innovation https://www.perplexity.ai/hub/blog/bullying-is-not-innovation

Read: Amazon’s Cease and Desist letter to Perplexity https://www.aboutamazon.com/news/company-news/amazon-perplexity-comet-statement

Read: Amazon’s Cease and Desist letter to Perplexity https://www.aboutamazon.com/news/company-news/amazon-perplexity-comet-statement

Shopify reports AI shopping traffic jumped 7x and orders 11x since January
The e-commerce platform’s dramatic growth in AI-driven commerce signals mainstream adoption of AI shopping assistants, with 64% of shoppers now likely to use AI for purchases. Shopify has partnered with ChatGPT, Perplexity, and Microsoft Copilot to embed shopping directly into AI conversations, positioning itself as the infrastructure layer for what it calls “agentic commerce.” The company’s access to data from millions of merchants gives it a competitive advantage as AI transforms how people discover and buy products online.

Shopify says AI traffic is up 7x since January, AI-driven orders are up 11x | TechCrunch https://techcrunch.com/2025/11/04/shopify-says-ai-traffic-is-up-7x-since-january-ai-driven-orders-are-up-11x/

Tesla shareholders approve Musk’s unprecedented $1 trillion compensation package
Tesla shareholders voted to approve Elon Musk’s massive pay package, potentially making him the world’s first trillionaire if the company hits ambitious performance targets. This represents the largest executive compensation deal in corporate history, tying Musk’s wealth directly to Tesla’s stock performance and market valuation milestones. The approval signals extraordinary shareholder confidence in Musk’s leadership despite ongoing concerns about his divided attention across multiple companies.

Tesla shareholders approve $1 trillion pay package for Musk | CNN Business https://edition.cnn.com/2025/11/06/business/musk-trillion-dollar-pay-package-vote

Tesla Shareholders Approve Elon Musk’s $1 Trillion Pay Package – WSJ https://www.wsj.com/business/autos/elon-musk-tesla-pay-package-vote-9abd5a73?st=d8Surv&reflink=desktopwebshare_permalink&mod=tldr

Musk predicts AI will eliminate digital jobs while sparing physical work
Musk called AI a “supersonic tsunami” that will destroy office jobs “like lightning” but leave cooking, plumbing, and electrical work intact longer. He believes AI combined with robotics will eventually make all human work optional, distinguishing between immediate white-collar disruption and delayed blue-collar impact.

Elon Musk describes AI as a “”supersonic tsunami”” that will overtake digital jobs “”like lightning.”” He expects physical jobs, like cooking, plumbing, and electrical work, to exist for much longer. He predicts that ultimately AI and robots combined will make working optional. https://x.com/TheHumanoidHub/status/1985404620391514377

Google launches Ironwood TPU with 10x performance boost over previous generation
Google’s new seventh-generation tensor processing unit delivers unprecedented AI computing power, claiming 10 times better peak performance than its TPU v5p and over 4 times improvement per chip versus the current Trillium generation. The Ironwood chips will power Google’s own frontier AI models for both training and deployment, potentially accelerating the development of more capable AI systems. This represents one of the largest single-generation performance leaps in specialized AI hardware.

Our 7th gen TPU Ironwood is coming to GA!  It’s our most powerful TPU yet: 10X peak performance improvement vs. TPU v5p, and more than 4X better performance per chip for both training + inference workloads vs. TPU v6e (Trillium). We use TPUs to train + serve our own frontier https://x.com/sundarpichai/status/1986463934543765973

Deutsche Telekom and Nvidia launch $1.1 billion Munich datacenter project
The facility will house 10,000 advanced GPUs including Nvidia’s latest DGX B200 systems, representing one of Europe’s largest AI infrastructure investments and signaling the continent’s push to build domestic AI computing capacity rather than rely on US cloud providers.

Telekom and NVIDIA building a $1.1B datacenter in Munich with 10k GPUs including DGX B200 and RTX PRO Servers https://x.com/scaling01/status/1985741851991621712

OpenAI’s finance chief backs federal insurance for chip investments
OpenAI’s CFO endorsed government-backed insurance for semiconductor investments, signaling that even leading AI companies see chip supply as a critical vulnerability. This represents a shift from typical tech industry resistance to government intervention, highlighting how chip shortages and geopolitical tensions around Taiwan have made hardware access an existential concern for AI development.

OpenAI CFO Would Support Federal Backstop for Chip Investments https://www.wsj.com/video/openai-cfo-would-support-federal-backstop-for-chip-investments/4F6C864C-7332-448B-A9B4-66C321E60FE7

OpenAI denies seeking government guarantees for massive datacenter investments
OpenAI publicly rejected claims that it wants taxpayer-backed guarantees for its datacenter projects, stating governments shouldn’t pick winners or bail out companies with poor business decisions. This clarification comes amid growing scrutiny of AI companies seeking public support for expensive infrastructure while positioning themselves as private market leaders.

I would like to clarify a few things. First, the obvious one: we do not have or want government guarantees for OpenAI datacenters. We believe that governments should not pick winners or losers, and that taxpayers should not bail out companies that make bad business decisions or”” / X https://x.com/sama/status/1986514377470845007

Michigan lands OpenAI’s largest-ever data center investment worth billions
OpenAI’s Stargate facility will be the state’s biggest single investment in history, creating 4,450 jobs and consuming over one gigawatt of power. The project distinguishes itself by using closed-loop water systems to avoid drawing from the Great Lakes and preserving 700 acres of natural land. Michigan beat other locations by offering workforce proximity to major universities and recent legislation reducing high-tech infrastructure costs.

Gigawatt-scale Stargate data center is the largest single investment in Michigan history: https://x.com/gdb/status/1984394938453528684

Whitmer announces largest-ever investment in Michigan with OpenAI data center | wzzm13.com https://www.wzzm13.com/article/news/local/michigan-lands-largest-economic-project-in-state-history-ai-facility/69-4abd2b41-9a11-4071-a589-d7fcc3a0a8a1

Quantinuum launches Helios quantum computer with 98 high-fidelity qubits
Quantinuum unveiled Helios, claiming it as the world’s most accurate quantum computer with 98 physical qubits and 48 error-corrected logical qubits. The system represents a significant leap in quantum computing reliability, as error-corrected logical qubits are essential for practical quantum applications but extremely difficult to achieve. This milestone could accelerate the timeline for quantum computers to solve real-world problems in chemistry, cryptography, and optimization that are impossible for classical computers.

Today, we launched Helios, a technological marvel redefining the possible. Helios is the most accurate quantum computer in the world, with 98 of the highest fidelity physical qubits ever released, and 48 error-corrected logical qubits. Learn more: https://x.com/QuantinuumQC/status/1986172816241402189

Introducing Helios: The Most Accurate Quantum Computer in the World https://www.quantinuum.com/blog/introducing-helios-the-most-accurate-quantum-computer-in-the-world

Coca-Cola creates holiday ad entirely using AI video generation tools
The beverage giant replaced its traditional film production with AI-generated footage for its iconic “Holidays are Coming” Christmas commercial, marking the first major brand to fully embrace AI video creation for a flagship campaign. This signals a potential industry shift toward AI-replacing traditional advertising production, though the move has sparked debate about creativity and job displacement in the creative sector. The ad maintains Coca-Cola’s familiar holiday imagery while demonstrating how AI video tools have reached commercial-grade quality for major brand campaigns.

Coca-Cola | Holidays are Coming, Fantastical :90 – YouTube https://www.youtube.com/watch?v=eoXX905YK6M

Google’s Veo 3.1 adds camera control to existing AI videos
Google’s video AI now lets users adjust camera angles and movement in previously generated clips, marking a shift from one-shot creation to iterative video editing. This represents the first major tool allowing post-generation modification of AI video content, potentially transforming how creators work with synthetic media by enabling refinement rather than starting over.

Veo 3.1 now has a Camera Adjustment Feature, allowing you to change the angle and movement of a previously generated video. Taking it out for a test spin, here’s our “”Test”” video, in the thread we’ll check out how the feature does! https://x.com/TheoMediaAI/status/1986104791454388289

OpenAI’s Sora creates hyper-realistic videos from simple text prompts
The AI video generator can produce Hollywood-quality footage including complex scenarios like Oscar scenes with boom mic mishaps, marking a breakthrough in AI’s ability to understand and visualize nuanced human storytelling. This represents a major leap beyond previous AI video tools that produced only basic, often distorted clips.

Sora: “”that infamous dramatic Oscar winning scene where the lead keeps getting hit by the boom mic but nobody notices”” https://x.com/emollick/status/1985923709786603845

Snapchat pays Perplexity $400 million to replace default AI search
Starting early 2026, Perplexity’s AI search engine will be integrated directly into Snapchat’s chat interface for all 477 million daily users, replacing Snap’s existing MyAI chatbot as the primary AI assistant. This marks one of the largest AI integration deals to date and represents a significant shift toward AI-powered search becoming native to social platforms. Snap’s stock jumped 9% on the announcement, with the company expecting the revenue to boost its bottom line starting in 2026.

Perplexity will be the default AI for all Snapchat users in the Snapchat app from January 2026. Excited to work together with Evan and Snap team on many more projects.”” / X https://x.com/AravSrinivas/status/1986205740273725686

Snap and Perplexity sign $400 million deal to put AI search directly in Snapchat https://www.engadget.com/social-media/snap-and-perplexity-sign-400-million-deal-to-put-ai-search-directly-in-snapchat-221101734.html

Snap Inc. – Snap and Perplexity Partner to Bring Conversational AI Search to Snapchat https://investor.snap.com/news/news-details/2025/Snap-and-Perplexity-Partner-to-Bring-Conversational-AI-Search-to-Snapchat/default.aspx

Snap stock jumps 9% after announcing Perplexity deal https://www.cnbc.com/2025/11/06/snap-stock-perplexity-ai-earnings.html

Snapchat + @perplexity_ai 🤝 Starting in early 2026, you’ll be able to ask questions, explore new ideas, and get credible answers right inside chat. AI that feels more personal, social, and fun! https://x.com/Snap/status/1986191838529601835

Starting in January, Perplexity will be the default AI for all Snapchat users in the Snapchat app. https://x.com/perplexity_ai/status/1986203714471010738

Researchers create humanoid robot that tracks complex motions despite real-world disruptions
A team from Tsinghua and Peking universities developed Any2Track, a system that enables humanoid robots to perform dancing, fighting, and athletic moves while adapting to terrain changes, external forces, and equipment variations. The breakthrough addresses a key robotics challenge by combining motion tracking with real-time adaptation, successfully transferring from simulation to a physical Unitree G1 robot without additional training. This represents a significant step toward robots that can operate reliably in unpredictable real-world conditions rather than controlled laboratory settings.

Today we’re rolling out major upgrades that significantly improve the performance and overall outcomes of the Comet Assistant. Comet can now handle more complex, multi-site workflows while working across multiple tabs in parallel. https://x.com/perplexity_ai/status/1986499432410718598

Any2Track: Galbot’s whole-body motion tracking system. https://x.com/TheHumanoidHub/status/1985219089015664676

Track Any Motions under Any Disturbances https://zzk273.github.io/Any2Track/

Perplexity enables trillion-parameter AI models to run on Amazon’s cloud infrastructure
Perplexity developed specialized software kernels that allow massive AI models like the 1-trillion parameter Kimi-K2 to run efficiently across multiple cloud servers connected by Amazon’s networking hardware. This breakthrough makes the largest open-source AI models accessible on standard cloud platforms for the first time, achieving competitive speeds despite Amazon’s network limitations compared to specialized AI hardware. The company published their first research paper demonstrating these kernels outperform existing solutions and enable viable deployment of models too large for single machines.

Enabling Trillion-Parameter Models on AWS EFA https://research.perplexity.ai/articles/enabling-trillion-parameter-models-on-aws-efa

Perplexity is the first to develop custom Mixture-of-Experts (MoE) kernels that make trillion-parameter models available with cloud platform portability. Our team has published this work on arXiv as Perplexity’s first research paper. Read more: https://x.com/perplexity_ai/status/1986101355896098836

NEO seized every single headline. https://x.com/TheHumanoidHub/status/1985429433327722609

NEO Home Robot https://www.1x.tech/neo

The world’s perception of 1X’s product launch is fascinating. Some thoughts: * 1X’s product is to give you 3 to 4 hours of in-home labor that does some chores well and more chores to an okay level, but in a completely safe and compliant manner. * At $500/month. For 120 hours a”” / X https://x.com/soumithchintala/status/1985391663712207056

NVIDIA CEO’s son joins company to lead humanoid robot development
Spencer Huang, Jensen Huang’s son, has become a product lead at NVIDIA Robotics, focusing on simulation frameworks and robot autonomy that could accelerate practical humanoid applications. This represents NVIDIA’s deeper push into physical AI beyond just providing chips, with the company now directly shaping how robots learn through synthetic data and reinforcement learning. The hiring signals NVIDIA’s belief that robotics represents the next major AI frontier after language models.

Spencer Huang of NVIDIA shares the first use cases he would love to see for humanoid robots https://x.com/TheHumanoidHub/status/1985767748358849019

We will be hearing more and more about Spencer Huang from now on. He recently joined NVIDIA, the family business led by his father Jensen Huang, and is already shaping the company’s vision for robotics and physical AI. His mix of curiosity, technical depth, and quiet confidence https://x.com/TheTuringPost/status/1985427046013813093

New episode: Spencer Huang is a Product Lead for @NVIDIARobotics, focusing on open-source simulation frameworks, synthetic data generation, and robot autonomy. Spencer breaks down RL breakthroughs unlocked by hardware, open-source simulation, synthetic data flywheels, the https://x.com/TheHumanoidHub/status/1984641886230102217

Royal Surrey Hospital reaches 10,000 robotic surgeries milestone since 2009
The achievement highlights robotics becoming mainstream in NHS cancer care, with the hospital now performing over 1,500 robotic procedures annually compared to just three in its first year. This growth reflects broader NHS plans to expand robotic surgery from 70,000 operations in 2023/24 to half a million by 2035, making robot-assisted procedures the default for nine out of ten keyhole surgeries within a decade.

The Royal Surrey Hospital performs 10,000th robotic surgery https://www.bbc.com/news/articles/c7v8176z7dlo

Robotic AI models now follow same scaling laws as ChatGPT
Researchers proved that robot foundation models improve predictably with more data and computing power, just like language models do, overturning previous failed attempts that used insufficient training data. This breakthrough suggests robots could achieve dramatic capability gains through the same scaling approach that made ChatGPT possible.

This was a key figure from the original GPT-3 paper. It proved LLMs could scale. We now believe we’re the first to see this hold true for robotic foundation models as well. It wasn’t obvious this would be true. We believe prior null results were due to insufficient useful data https://x.com/E0M/status/1985766175255773483

Google prepares Nano Banana 2 image generation model for next week
Google is readying GEMPIX 2, the second version of its Nano Banana image generation model, with announcement cards appearing in Gemini’s interface signaling an imminent launch. The model targets creators and professionals using AI for image generation, following the success of the original Nano Banana which helped grow Gemini’s user base and market share. Pre-release UI changes typically appear about a week before public rollout, suggesting availability as soon as next week.

Google is preparing Nano Banana 2 for the upcoming release https://www.testingcatalog.com/google-is-preparing-nano-banana-2-for-the-upcoming-release/

ChatGPT develops unexpected humor in creative image generation tasks
OpenAI’s chatbot now produces genuinely amusing visual concepts when asked for absurd scenarios, suggesting AI systems are beginning to understand and generate comedy beyond simple pattern matching. This represents a notable shift from earlier versions that struggled with humor’s contextual and cultural nuances, indicating progress in AI’s grasp of human creativity and wit.

Same prompt a year later: “give me images of those bags that hold different cheap costumes from Halloween stores, but make the costumes really weird” ChatGPT is actually getting close to funny at times. https://x.com/emollick/status/1984363186896924684

Apple plans to use Google’s massive 1.2 trillion parameter Gemini model for Siri
This partnership would give Apple access to one of the world’s largest AI models, potentially transforming Siri from a basic voice assistant into a sophisticated conversational AI. The 1.2 trillion parameter size puts Gemini among the most powerful language models available, suggesting Apple is prioritizing AI capability over developing its own competing technology. This represents a significant strategic shift for Apple, which typically builds core technologies in-house.

Apple Plans to Use 1.2 Trillion Parameter Google Gemini Model to Power New Siri – Bloomberg https://www.bloomberg.com/news/articles/2025-11-05/apple-plans-to-use-1-2-trillion-parameter-google-gemini-model-to-power-new-siri

Apple just leaked the size of Gemini 3 Pro – 1.2T params https://x.com/scaling01/status/1986158792128508218

Ilya Sutskever reveals details behind OpenAI’s failed coup attempt
Former OpenAI chief scientist Ilya Sutskever testified he spent over a year planning Sam Altman’s ouster, submitting a 52-page memo accusing the CEO of “lying, undermining executives, and pitting them against each other.” The November 2023 firing lasted only four days after 738 employees threatened to quit, but depositions reveal OpenAI seriously considered merging with Anthropic during the chaos. This marks the first detailed insider account of the power struggle that nearly toppled the world’s most valuable AI company.

** HIGHLY CONFIDENTIAL ** 13 Videotaped Deposition of ILYA SUTSKEVERgov.uscourts.cand.433688.340.1.pdf https://storage.courtlistener.com/recap/gov.uscourts.cand.433688/gov.uscourts.cand.433688.340.1.pdf

Lots of RTs on my Helen Toner interview today. Over a year later, we hear Ilya’s side of this story. The Brockman beef was unexpected (Ilya wrote a memo on him too), as was the failed Anthropic merger. Interesting that most screenshots in Ilya’s 52 page memo were sourced from”” / X https://x.com/bilawalsidhu/status/1985253917106827682

Former OpenAI Exec Explains Why He Tried to Do a Coup Against Sam Altman https://gizmodo.com/former-openai-exec-explains-why-he-tried-to-do-a-coup-against-sam-altman-2000680769

Amazon launches Chronos-2 to predict any data pattern without training
The new forecasting model can analyze any type of data trend immediately without needing specific training for each use case, potentially transforming how businesses predict sales, demand, and market changes across industries.

🤖 From this week’s issue: Amazon introduced Chronos-2, a foundation model designed to handle arbitrary forecasting tasks — univariate, multivariate, and covariate informed — in a zero-shot manner. https://x.com/dl_weekly/status/1985346603108991015

Anthropic’s code execution approach cuts AI agent token usage by 98.7%
Anthropic’s new Model Context Protocol technique lets AI agents access thousands of tools through code rather than direct calls, reducing token consumption from 150,000 to 2,000 tokens per complex task. This solves a major scaling problem where agents loading many tool definitions upfront were becoming prohibitively expensive and slow. The approach transforms how agents interact with external systems like Google Drive and Salesforce, making enterprise AI workflows dramatically more cost-effective.

Code execution with MCP: building more efficient AI agents \ Anthropic https://www.anthropic.com/engineering/code-execution-with-mcp

Anthropic just posted another banger guide. This one is on building more efficient agents to handle more tools and efficient token usage. This is a must-read for AI devs! (bookmark it) It helps with three major issues in AI agent tool calling: token costs, latency, and tool https://x.com/omarsar0/status/1986099467914023194

ElevenLabs CEO predicts voice will replace screens as primary AI interface
ElevenLabs’ chief executive argues that voice interaction will become the dominant way people communicate with AI systems, moving beyond today’s text-based chatbots and visual interfaces. This shift matters because it could fundamentally change how we access information and services, making AI more accessible to people who struggle with traditional computing while enabling hands-free interaction in cars, homes, and workplaces. The company’s focus on advanced voice synthesis technology positions it at the center of this potential transformation from typing to talking with machines.

(32) ElevenLabs CEO: Why Voice is the Next AI Interface – YouTube https://www.youtube.com/watch?v=ZqCEHR4wjxg

Google Gemini can now search your personal emails and documents for research reports
Google integrated its Deep Research feature with Workspace apps, letting the AI analyze your Gmail, Drive files, and chats to create comprehensive reports that combine private data with web research. This marks a shift toward AI assistants that can access and synthesize personal information rather than just public data. The feature launches on desktop first, with mobile access coming soon.

One of our most requested features is here: Gemini Deep Research can now draw on context from @GoogleWorkspace — meaning you can create even more comprehensive reports by pulling in information directly from your Gmail, Drive (including Docs, Slides, Sheets and PDFs) and Google https://x.com/Google/status/1986190599150518573

Google Gemini’s Deep Research can look into your emails, drive, and chats | The Verge https://www.theverge.com/ai-artificial-intelligence/814878/google-ai-gemini-deep-research-personalized

Introducing the File Search Tool in Gemini API https://blog.google/technology/developers/file-search-gemini-api/

Now, Gemini’s Deep Research can pull in info from @Gmail, @GoogleDrive, and Chat when you connect your @GoogleWorkspace account to give you more context-aware reports. To try it, just select “Deep Research” in Gemini on desktop and choose your sources. Coming to mobile soon.”” / X https://x.com/GeminiApp/status/1986472318873555058

We’ve launched the File Search Tool, a fully managed RAG system integrated into the Gemini API that simplifies grounding models with your private data to deliver more accurate, verifiable responses. – $0.15/m tokens for indexing, free storage and embedding generation at query https://x.com/_philschmid/status/1986506204240347520

Google Maps integrates Gemini for hands-free voice navigation and landmark guidance
Google is adding conversational AI to Maps, letting drivers ask complex questions like finding vegan restaurants with parking details using voice commands, while also providing directions based on visible landmarks like “turn right after the Thai restaurant” instead of just distances. This marks the first major AI assistant integration into turn-by-turn navigation, potentially changing how millions navigate daily by making driving safer and more intuitive.

Google Maps is getting a powerful boost with Gemini, making navigation smarter and easier. ✨ Learn more about the new features ↓”” / X https://x.com/Google/status/1986164830588248463

Google Maps launches Gemini features, including landmark navigation https://blog.google/products/maps/gemini-navigation-features-landmark-lens/

Gemini has arrived as your hands-free driving assistant in the @GoogleMaps app. Find places along your route, check for EV availability, and share your ETA just by asking.  Gemini can also help with multi-step tasks like “find me a restaurant that serves vegetarian tacos within https://x.com/sundarpichai/status/1986119293914792338

AI agents now follow “rule of two” security requiring dual authorization
Researchers developed a security framework where AI agents must get approval from two independent sources before taking high-risk actions, addressing growing concerns about autonomous AI systems making dangerous decisions without human oversight. Early tests show the approach significantly reduces harmful outputs while maintaining agent effectiveness.

Agents Rule of Two: A Practical Approach to AI Agent Security https://ai.meta.com/blog/practical-ai-agent-security/

Microsoft adds app building and workflow automation to Copilot chat
Microsoft integrated app creation and workflow automation directly into its Copilot chat interface, allowing users to build functional applications and automate business processes through conversational commands within their existing Microsoft 365 environment. This represents a significant shift from traditional development requiring coding skills to natural language app creation, potentially democratizing software development for millions of office workers.

Super energized about this! With our App Builder and Workflow agents, you can now build apps and automate workflows in minutes, right in M365 Copilot chat. Here’s an example. https://x.com/satyanadella/status/1983372239052710047

ChatGPT launches agent mode to autonomously complete tasks for users
OpenAI’s new agent mode allows ChatGPT to independently research, plan, and execute multi-step tasks while users continue browsing, marking a shift from conversational AI to autonomous task completion. This capability is currently available in preview for paying subscribers across Plus, Pro, and Business tiers. The feature represents a significant evolution beyond simple chat interactions toward AI systems that can operate independently to achieve user goals.

Turn on agent mode and ChatGPT can take action for you—research, plan, and get things done while you browse. Now in preview for Plus, Pro, and Business users. https://x.com/OpenAI/status/1984304194837528864

OpenAI launches Aardvark agent to automatically find security vulnerabilities
The AI agent uses GPT-5 to identify and fix software security bugs without human intervention, marking a shift from AI as a coding assistant to autonomous security researcher. Currently in private beta testing, this represents one of the first specialized AI agents designed to handle complex cybersecurity tasks end-to-end.

A new security agent called Aardvark:”” / X https://x.com/sama/status/1984002552158154905

Introducing Aardvark: OpenAI’s agentic security researcher | OpenAI https://openai.com/index/introducing-aardvark/

Now in private beta: Aardvark, an agent that finds and fixes security bugs using GPT-5. https://x.com/OpenAI/status/1983956431360659467

Whisper loses its crown as top open-source speech transcription model
OpenAI’s once-dominant Whisper model has been overtaken by newer open-source alternatives that achieve better word error rates in transcription accuracy tests. This marks a significant shift in the speech-to-text landscape, showing how rapidly AI capabilities are advancing beyond even well-established models from major tech companies.

Whisper no longer wears the open weights transcription accuracy crown with new entrants achieving better Artificial Analysis Word Error Rate scores Once considered the default choice for open weights transcription, OpenAI’s Whisper has now been surpassed by newer open weights https://x.com/ArtificialAnlys/status/1986100695989145649

PayPal launches AI commerce services targeting small business merchants
PayPal introduced “agent ready” payments and “store sync” catalog tools that let existing merchants sell through AI platforms like Perplexity without technical changes. The move positions PayPal to capture transaction fees as shopping shifts from traditional websites to AI-powered discovery, with partnerships across major e-commerce platforms making millions of merchants instantly compatible with conversational AI shopping.

Commerce is entering a new era—powered by AI. Introducing PayPal’s agentic commerce services, helping merchants—especially small businesses—thrive in AI-driven shopping. Includes: 💳 agent ready 🛍️ store sync Learn more 👉 https://x.com/PayPal/status/1983633727138812327

Press Release: PayPal Launches Agentic Commerce Services to Power AI-Driven Shopping – Oct 28, 2025 https://newsroom.paypal-corp.com/2025-10-28-PayPal-Launches-Agentic-Commerce-Services-to-Power-AI-Driven-Shopping

AI system makes seven validated scientific discoveries across multiple fields
Sakana AI’s “AI Scientist” represents a breakthrough in autonomous research, running independently for days to generate genuine scientific findings that external experts have confirmed, marking a shift from AI as a research tool to AI as an actual researcher.

After two years of work, we’ve made an AI Scientist that runs for days and makes genuine discoveries. Working with external collaborators, we report seven externally validated discoveries across multiple fields. It is available right now for anyone to use. 1/5 https://x.com/andrewwhite01/status/1986094948048093389

Edison Scientific launches AI scientist that reads 1500 papers per run
The new Kosmos system processes vastly more information than previous AI research tools, with beta users estimating it completes six months of work in one day and achieving 79.4% accuracy in conclusions across seven validated scientific discoveries.

Edison Scientific (a brand-new company spun out of FutureHouse) releases Kosmos: An AI Scientist for Autonomous Discovery “”Our beta users estimate that Kosmos can do in one day what would take them 6 months, and we find that 79.4% of its conclusions are accurate.”” The paper https://x.com/iScienceLuvr/status/1986023952037417109

Kosmos: An AI Scientist for Autonomous Discovery https://edisonscientific.com/articles/announcing-kosmos

A year ago, I would not have expected the first academic field to seem to reach a consensus that AIs will accelerate research (which is not the same thing as autonomous research) would be math But that appears to be happening based on math professors in my feed and elsewhere.”” / X https://x.com/emollick/status/1984388281061282081

Chinese AI lab Moonshot releases Kimi K2 Thinking model
Moonshot AI released Kimi K2 Thinking, a 1-trillion parameter open-source reasoning model that matches or exceeds closed models like GPT-4 on key benchmarks including coding tasks. The model can execute 200-300 sequential tool calls autonomously and achieved 71% on SWE-Bench Verified, marking what experts call the closest open models have come to frontier performance. This represents a significant shift as Chinese AI labs increasingly challenge American closed-source dominance through rapid open releases.

@Kimi_Moonshot Congratulations to the entire Moonshot team — today is a great day for open source everywhere. We’re excited to continue supporting Kimi models with fast inference on Baseten. https://x.com/basetenco/status/1986494013109903362

@QuixiAI @Kimi_Moonshot a single H200 node is enough😃”” / X https://x.com/vllm_project/status/1986626058897269070

📢 New Model(s) Drop: Kimi K2 Thinking and Kimi K2 Thinking Turbo are now on Yupp! This pair of thinking models from @Kimi_Moonshot specialize in deep reasoning tasks. We explored their capabilities with some prompts on Yupp: https://x.com/yupp_ai/status/1986469027997491422

🚀 Hello, Kimi K2 Thinking! The Open-Source Thinking Agent Model is here. 🔹 SOTA on HLE (44.9%) and BrowseComp (60.2%) 🔹 Executes up to 200 – 300 sequential tool calls without human interference 🔹 Excels in reasoning, agentic search, and coding 🔹 256K context window Built https://x.com/Kimi_Moonshot/status/1986449512538513505

🚨 New Open Source Model Update! Touted for its reasoning and coding strengths, Kimi K2 Thinking by @Kimi_Moonshot is now live for both Text and WebDev in Battle, Side by Side and Direct. Bring your toughest prompts! 💪 The last time Kimi K2 was in the Arena with a new model, https://x.com/arena/status/1986482438768673107

5 Thoughts on Kimi K2 Thinking – by Nathan Lambert https://www.interconnects.ai/p/kimi-k2-thinking-what-it-means

70% on SWE bench verified 30% terminal bench those are two intuitive thresholds for “”actually useful and not frustrating”” coding assistant. Kimi k2 thinking got 71.3% on SWE-Bench Verified 47.1% on Terminal-Bench”” / X https://x.com/andrew_n_carr/status/1986538323876454461

Congrats to the Kimi K2 team on the great numbers on our SWE-bench Verified, SWE-bench Multilingual and SciCode benchmarks!! https://x.com/OfirPress/status/1986475891158040760

Kimi AI – Kimi K2 is Live https://www.kimi.com/

Kimi API is barely alive right now kinda slow ~20tks/s and get quite a few timeouts / network errors when I let the model reason for a long time”” / X https://x.com/scaling01/status/1986476278908920061

Kimi K2 Thinking feels like a big milestone for open-source AI. The first time in a while that open-source gets ahead of proprietary APIs on their big area of focus (agents). Fun to see that it’s happening at a time when the proprietary APIs have the most money/attention”” / X https://x.com/ClementDelangue/status/1986833436607160600

Kimi K2 Thinking https://moonshotai.github.io/Kimi-K2/thinking.html

Kimi K2 Thinking is now available in anycoder https://x.com/_akhaliq/status/1986468663600337125

Kimi K2 Thinking is the new leading open weights model: it demonstrates particular strength in agentic contexts but is very verbose, generating the most tokens of any model in completing our Intelligence Index evals @Kimi_Moonshot’s Kimi K2 Thinking achieves a 67 in the https://x.com/ArtificialAnlys/status/1986911675820446013

Kimi K2 Thinking just launched on Product Hunt! 🥳 Not chasing votes, just using PH as a clean milestone log for our model updates. 🙂 Huge thanks to the helpful team from @ProductHunt https://x.com/crystalsssup/status/1986714377983304137

Kimi-K2 is an exceptional base model GPQA Diamond 77% GPT-4.5 only got 71.4%”” / X https://x.com/scaling01/status/1986112227875954967

Kimi-K2 Reasoning is coming very soon just got merged into VLLM LETS FUCKING GOOOO im so hyped im so hyped im so hyped https://x.com/scaling01/status/1986071916541870399

Kimi-K2 reasoning is landing soon; it just got merged into vLLM https://x.com/cedric_chee/status/1986073808672067725

Kimi-K2 Thinking ranking 19th on SimpleBench improving Kimi-K2s score from 26.3% (rank 33) to 39.6% This makes it the 3rd best open-source model on SimpleBench. Other chinese open-source models like DeepSeek R1 0528 and DeepSeek V3.1 beat it by roughly 1 %. https://x.com/scaling01/status/1986846212050362510

Live in Cline: kimi-k2-thinking https://x.com/cline/status/1986512739490275680

MoonshotAI has released Kimi K2 Thinking, a new reasoning variant of Kimi K2 that achieves #1 in the Tau2 Bench Telecom agentic benchmark and is potentially the new leading open weights model Kimi K2 Thinking is one of the largest open weights models ever, at 1T total parameters https://x.com/ArtificialAnlys/status/1986541785511043536

moonshotai/Kimi-K2-Thinking · Hugging Face https://huggingface.co/moonshotai/Kimi-K2-Thinking

ollama run kimi-k2-thinking:cloud Kimi K2 Thinking is Moonshot AI’s best open-source thinking model. Try it on Ollama’s cloud! https://x.com/ollama/status/1986640693108863271

Our first research paper: custom Mixture-of-Experts (MoE) kernels that make deployment of trillion-parameter models like Kimi K2 viable for the first time on AWS EFA https://x.com/AravSrinivas/status/1986106660386222592

Unsurprisingly, Kimi K2 Thinking is already number one trending on HF. The AI frontier is open-source! https://x.com/ClementDelangue/status/1986827413532057712

🚀 Day 0 support: Kimi K2 Thinking now running on vLLM! In partnership with @Kimi_Moonshot, we’re proud to deliver official support for the state-of-the-art open thinking model with 1T params, 32B active. Easy deploy in vLLM (nightly version) with OpenAI-compatible API: What https://x.com/vllm_project/status/1986455911066706160

It even compares with GPT 5 Pro on some benches. Looks like Kimi’s interpretation of Pro mode is 8 samples + self reflection https://x.com/nrehiew_/status/1986453238552666320

From my tests, Kimi K2 thinking is better than everything Xai, Anthropic, Google has to offer atm. The only thing that is better than this is Gpt 5 codex (at code) and Gpt 5 pro (at high level algorithm design) It beats the SOTA at creative writing by a mile. Good work”” / X https://x.com/karmay007/status/1986454592809529493

Leave a Reply

Trending

Discover more from Ethan B. Holland

Subscribe now to keep reading and get access to the full archive.

Continue reading