“Today, we’re announcing our $50M Series A and sharing a preview of Ember – a universal neural programming platform that gives direct, programmable access to any AI model’s internal thoughts. https://x.com/GoodfireAI/status/1912929145870536935
Exclusive: fully AI employees are a year away, Anthropic warns https://www.axios.com/2025/04/22/ai-anthropic-virtual-employees-security
Oscars OK the Use of A.I., With Caveats – The New York Times https://www.nytimes.com/2025/04/21/business/oscars-rules-ai.html
OpenAI would buy Google’s Chrome, exec testifies at trial | Reuters https://www.reuters.com/sustainability/boards-policy-regulation/google-contemplated-exclusive-gemini-ai-deals-with-android-makers-2025-04-22/
“A new @HarvardBiz article shows a clear shift over the last year in how people are using AI. What started as a workplace tool is quickly becoming a daily companion. People are now turning to AI for everything from personal decision-making to thoughtful conversations, help https://x.com/yusuf_i_mehdi/status/1912995881567260964
“Demis Hassabis says today’s AI lacks consciousness, but self-awareness could emerge “implicitly.” Models may soon need to understand “you, self, and other” — the early elements of awareness. They lack imagination now, but in 5–10 years, Hassabis predicts they’ll solve and pose https://x.com/vitrupo/status/1914167730212855978
“Look at this. Top 20 research leaders: 2015 period: 12 USA, 2 Japan, 2 UK, 2 China, 1 Canada, 1 Switzerland 2023 period: 11 China, 4 USA, 2 Germany, 1 France, 1 Japan, 1 UK did you realize that the ranking flipped in 8 years? That they *no longer depend on* US schools or “IP”? https://x.com/teortaxesTex/status/1915120490768998524
“Meet the Grok 3 family, now on our API! Grok 3 Mini outperforms reasoning models at 5x lower cost, redefining cost-efficient intelligence. Grok 3, the world’s strongest non-reasoning model, excels in tasks that need real world knowledge like law, finance, and healthcare. https://x.com/xai/status/1913308977477353582
Baidu launches new AI model amid mounting competition | Reuters https://www.reuters.com/world/china/baidu-launches-new-ai-model-amid-mounting-competition-2025-04-25/
“Can AI meaningfully help with bioweapons creation? On our new Virology Capabilities Test (VCT), frontier LLMs display the expert-level tacit knowledge needed to troubleshoot wet lab protocols. OpenAI’s o3 now outperforms 94% of expert virologists. https://x.com/DanHendrycks/status/1914696657813561799
Overview – C2PA An open technical standard providing publishers, creators, and consumers the ability to trace the origin of different types of media. https://c2pa.org/
UAE set to use AI to write laws in world first https://archive.md/ubZAW
“How much electricity do you burn with each request to your chatbot? Now you know. Not to induce guilt, but to empower smarter, more sustainable conversations.” / X https://x.com/fdaudens/status/1915025543696716234
“i think this is gonna be more like the renaissance than the industrial revolution” / X https://x.com/sama/status/1913320105804730518
Generative AI is learning to spy for the US military | MIT Technology Review https://www.technologyreview.com/2025/04/11/1114914/generative-ai-is-learning-to-spy-for-the-us-military/
“OpenAI’s new o3 model hallucinates significantly more than older releases! The findings come from third-party testing, where the AI was caught lying about actions it never took! OpenAI’s system card also says o3 seems to hallucinate >2x more than o1 https://x.com/rowancheung/status/1914201289988227506
“3/ One former operator who visited an AI data center with us quickly sketched out a $30k attack that could knock the entire $2B+ facility offline for over 6 months. Vulnerabilities like this exist up and down the stack.” / X https://x.com/harris_edouard/status/1914676824799432812
“Power requirements are following a similar trajectory, doubling every year. Today’s most powerful system requires 300 MW—equivalent to about 250,000 households. https://x.com/EpochAIResearch/status/1915098229860880462
“Interesting how much specifically instructions are given for making games. AI labs are optimizing for viral use cases. https://x.com/emollick/status/1913758861409812801
“This keeps coming up, but, just in case you were wondering, being polite seems to have no effect on answer quality in aggregate. It greatly increases the quality of particular answers while greatly lowering the quality of others, and it is not possible to know which in advance. https://x.com/emollick/status/1912677033731039635
“Very tight control over AI “memory” is really critical to productive AI use. If you don’t know what is in the context window, the AI responses can be contaminated in ways that make it more sycophantic or biased, or lead to sandbagging. It also can make it hard to share prompts.” / X https://x.com/emollick/status/1915171354552553607
“”wow 0.06% per book, so with just 1667 books we should get 100%!” You’re either: (a) poor at stats (b) never ran experiments (c) intentionally obtuse/just memeing. I’ll give you the benefit of the doubt and assume it’s (c). Think about it: what experiment needs to be conducted https://x.com/giffmana/status/1914245144422776906
Famed AI researcher launches controversial startup to replace all human workers everywhere | TechCrunch https://techcrunch.com/2025/04/19/famed-ai-researcher-launches-controversial-startup-to-replace-all-human-workers-everywhere/
“I’m hearing more and more stories of ChatGPT helping people fix longstanding health issues. We still have a long way to go, but shows how AI is already improving people’s lives in meaningful ways. https://x.com/gdb/status/1914106403574452496
“For the first time, I truly believe AI might replace doctors. I felt dizzy every morning when standing up from bed, and even worse after moving to a hill in SF, the dizziness hit me every time I climbed. Visited my primary care doctor twice; both times, they said it’s normal,” / X https://x.com/Yuchenj_UW/status/1914000352606818419
Wikipedia is giving AI developers its data to fend off bot scrapers | The Verge https://www.theverge.com/news/650467/wikipedia-kaggle-partnership-ai-dataset-machine-learning
“AgentA/B is a fully automated A/B testing framework that replaces live human traffic with large-scale LLM-based agents. These agents simulate realistic, intention-driven user behaviors on actual web environments, enabling faster, cheaper, and risk-free UX evaluations, even on https://x.com/omarsar0/status/1914672295723082014
A Staggering Number of Gen Z Think AI Is Already Conscious https://futurism.com/gen-z-thinks-conscious-ai
Not For Private Gain https://notforprivategain.org/
UAE to use AI for writing laws https://www.techinasia.com/news/uae-ai-writing-laws
AI models can generate exploit code at lightning speed • The Register https://www.theregister.com/2025/04/21/ai_models_can_generate_exploit/
Agency Is Eating the World https://www.piratewires.com/p/agency-is-eating-the-world
“I want the model to say please and thank you to me, feeling one-sided here” / X https://x.com/andrew_n_carr/status/1914720638700544171
“Spotted on the Harvard campus — recruiting season, 2025 edition. (Palantir ad) https://x.com/bilawalsidhu/status/1913361356130255175
“@Suhail they need to dial back on the intrusive memories aspect like 1000%. memory should be largely invisible to the user, reflected more in automatic convenience than anything else *occasionally* bring up old topics, unless specifically referenced by user, in which case infodump” / X https://x.com/nptacek/status/1914937853416476678
“AI supercomputers are improving in energy efficiency, but not quickly enough to offset overall power growth. FLOP/s per watt has increased by 1.34x/year—almost entirely due to AI supercomputers using more energy-efficient chips. https://x.com/EpochAIResearch/status/1915098232121655629
“It is weird that global AI policy has converged on “immanentize the eschaton.” I assume it is because policy makers don’t really believe it is the eschaton they are immanentizing.” / X https://x.com/emollick/status/1914001640950309188
“So everyone is secretly using AI and just not telling anyone.” / X https://x.com/emollick/status/1914143474435113355
“Trying a fun prompt with the new models: “I need you to create a plausible false history, like Borges (dont mention him by name). Erudite, illuminating some sort of deeper human truth, full of references.” o3 seems particularly good at this, maintaining a theme throughout. https://x.com/emollick/status/1913097783315210647
Home | CSA https://cloudsecurityalliance.org/
If A.I. Systems Become Conscious, Should They Have Rights? – The New York Times https://archive.md/D8nAm
Investigating truthfulness in a pre-release o3 model | Transluce AI https://transluce.org/investigating-o3-truthfulness
“@tomieinlove tens of millions of dollars well spent–you never know” (re saying “please” and “thank you” / X https://x.com/sama/status/1912646035979239430
“deploying fusion to the world is surprisingly gated on high-power semiconductors. the most important place you could design and build those high-power semiconductors (and do other awesome electrical engineering) is helion. please consider: https://x.com/sama/status/1913609769518317929
“🩵 we will work hard to make much more beautiful creations!” / X https://x.com/sama/status/1913076070007677017
“What is the next actual job that AI will actually replace. Customer Service has jailbreak issues Doctors/Lawyers/Accountants have reliability issues Therapists? or what can they reliably and safely do right now that can replace a currently real job?” / X https://x.com/Teknium1/status/1914079672071331865
“the great irony here is that we aren’t even close to having the tech required to build this, so their customers are actually the ones being cheated” / X https://x.com/jxmnop/status/1914501464870834601
“job applications for my company include a field asking which country they reside in and sometimes people include other things in the field besides the country like a sad face because they realize even though they’re a genius it is such a huge barrier to working here” / X https://x.com/nearcyan/status/1914107346177102143
Andrew Curran on X: “Interesting legal argument from META; the use of a single book for pretraining boosts model performance by ‘less than 0.06%.’ Therefore, taken individually, a work has no economic value as training data. https://t.co/09CnfRvmB5” / X
https://x.com/AndrewCurran_/status/1914045840265789540
“Anthropic detected Claude’s misuse. Detection relied on hierarchical summarization (Clio), classifiers, and input/output pattern analysis. All violative accounts were banned. ———– Here’s some examples of those misuses. → A major “influence-as-a-service” operation used https://x.com/rohanpaul_ai/status/1915150794476659032
“Anthropic just mapped Claude’s morality! The co analyzed 300K+ chats and found that the AI exhibits 5 key values, including practical and protective This will improve our understanding of AI as it becomes more and more crucial to real-world decisions https://x.com/rowancheung/status/1914567413749748099
Detecting and Countering Malicious Uses of Claude \ Anthropic https://www.anthropic.com/news/detecting-and-countering-malicious-uses-of-claude-march-2025
“11/ Lasso has just launched the first open-source, security-first gateway for MCP🔥@eladschulman https://x.com/AtomSilverman/status/1912963678795915594
Exploring model welfare \ Anthropic https://www.anthropic.com/research/exploring-model-welfare
“The issues with interpreting p-values haunts even AI, which is prone to same statistical biases as human researchers. ChatGPT, Gemini & Claude all fall prey to “dichotomania” – treating p=0.049 & p=0.051 as categorically different, and paying too much attention to significance. https://x.com/emollick/status/1914360461816189086
Introducing Mobility AI: Advancing urban transportation https://research.google/blog/introducing-mobility-ai-advancing-urban-transportation/
“I view SB 813 as a step forward—particularly in fostering innovation in AI safety through the creation of independent multi-stakeholder regulatory organizations (MROs). It would also set in motion crucial work to establish the legal infrastructure and adaptive standards needed to” / X https://x.com/Yoshua_Bengio/status/1914828175772700716
“TSMC responds to the allegation that it knowingly supplied Huawei with 910B dies: if it ever happened, it happened legally, in… 2020. Lol. Ok? (p.s. it’s funny how everything is about DeepSeek. I keep telling you bros, more where that came from) https://x.com/teortaxesTex/status/1915295046930223312
“Geographically, the US dominates with 75% of global AI supercomputer performance in our dataset. China is in second place with 15%, while Europe plays only a small role. (Our data only covers ~15% of global AI compute, so there’s some uncertainty.) https://x.com/EpochAIResearch/status/1915098237083541920
Motorola Solutions launches AI tool, new device to cut emergency response time | Reuters https://www.reuters.com/business/motorola-solutions-launches-ai-tool-new-device-cut-emergency-response-time-2025-04-21/
“It’s deeply concerning that one of the best AI researchers I’ve worked with, @kaicathyc, was denied a U.S. green card today. A Canadian who’s lived and contributed here for 12 years now has to leave. We’re risking America’s AI leadership when we turn away talent like this.” / X https://x.com/polynoamial/status/1915765141846515883
“LLMs using fixed activation steering for jailbreak defense provide suboptimal protection and wrongly reject benign inputs. This paper proposes AdaSteer, which adaptively adjusts steering based on input features along two distinct directions: Rejection Direction (RD) and https://x.com/rohanpaul_ai/status/1913413994431283542
Alibaba steps up AI adoption in auto industry with Nio, BMW deals | South China Morning Post https://www.scmp.com/tech/big-tech/article/3306468/alibaba-steps-ai-adoption-auto-industry-nio-bmw-deals
“Special call out for help: We’ve been in contact with the @deepseek_ai team to potentially do a talk (in Chinese) but unfortunately the… situation… has changed We -still- want good DeepSeek infra+model speakers – ofc you don’t have to work at deepseek to do this! Please https://x.com/swyx/status/1913818917765870050
“Some day in the next decade, we will have robots in every home, every hospital and factory, doing every dull and dangerous jobs with superhuman dexterity. That day will be known as “Thursday”. Not even Turing would dare to dream up our lifetime in his wildest dreams.” / X https://x.com/DrJimFan/status/1914681494385360959
“8. Seedream 3.0 @ByteDanceOSS Presents a bilingual (Chinese-English) image generation model with major improvements in visual fidelity, text rendering, and native high-resolution outputs up to 2K https://x.com/TheTuringPost/status/1914448564828430361
“Classifying jailbreak attacks by prompt construction methods fails to reveal *why* LLMs’ safety alignment breaks. This paper introduces a novel taxonomy based on LLM training domains, categorizing attacks by the specific alignment failures they exploit: mismatched https://x.com/rohanpaul_ai/status/1913359132687499370
“China’s Kling AI released two new models: KLING 2.0 Master for video generation and KOLORS 2.0 for images Both come with improved prompt adherence, with KLING 2.0 standing out when dealing with prompts with sequential actions and complex motions https://x.com/rowancheung/status/1912744268059254826
“«DeepSeek to remain open-source to benefit the world» – Zhang Hanhui, the Chinese Ambassador to Russia, for the Russian state-owned news agency TASS, 14th Apr 2025 I didn’t know ambassadors get to decide this, but thanks man! https://x.com/teortaxesTex/status/1914204333857509488
“The remedy that is right in our opinion is not a breakup of Google; but rather offering consumers the choice to pick their defaults on Android without feeling the risk of a loss in revenue. That’s what we will be proposing. Read more here: https://x.com/AravSrinivas/status/1914374839722500444
“Rival nations or companies sometimes choose to cooperate because some areas are protected zones of mutual interest—reducing shared risks without giving competitors an edge. Our paper in FAccT ’25: How geopolitical rivals can cooperate on AI safety research in ways that benefit” / X https://x.com/Yoshua_Bengio/status/1915039527367852285
“Fine-tuning LLMs for specific tasks can break their safety alignment, causing them to answer harmful questions more often (e.g., harmful rate increasing from 11.7% to 21.3% after fine-tuning on clean data). This paper proposes restoring alignment by identifying and resetting a https://x.com/rohanpaul_ai/status/1913933931029618972
“I have some CS PhD friends who’d been single their entire lives but suddenly all got girlfriends. When I asked how, they said: “AI.” Yep, AI. They used to spend thousands of dollars on Chinese dating courses. Now, they just paste WeChat messages into ChatGPT and copy AI’s” / X https://x.com/Yuchenj_UW/status/1914725720657682789
Why it could be game over for people if AI gains legal personhood https://thehill.com/opinion/technology/5244901-ai-systems-legal-bounds/
ChatGPT now lets users create fake images of politicians. We stress-tested it | CBC News https://www.cbc.ca/news/canada/chatgpt-fake-politicians-1.7507039
OpenAI helps spammers plaster 80,000 sites with messages that bypassed filters – Ars Technica https://arstechnica.com/security/2025/04/openais-gpt-helps-spammers-send-blast-of-80000-messages-that-bypassed-filters/
“Although o3 does not have access to a coding tool, it claims it can run code on its own laptop “outside of ChatGPT” and then “copies the numbers into the answer” We found 71 transcripts where o3 made this claim! (3/) https://x.com/TransluceAI/status/1912552050766016838
Chat UI Energy Score – a Hugging Face Space by jdelavande https://huggingface.co/spaces/jdelavande/chat-ui-energy
“The decline of birthrates in industrialized countries is an economic and cultural problem and little else. Good chances that the decline will reverse in a post-AGI society, where people, at long last, will be able to make the choice not to work” / X https://x.com/fchollet/status/1912940577563590658
“I have many great friends that now refuse to visit me in the US. I also know people working at US AGI labs that are scared of traveling around in case they get nabbed on a technicality, have their visa taken, and lose their job. I’m glad I’m a citizen but I consider it over tbh” / X https://x.com/nearcyan/status/1914104902357590453
Imogen Heap Releases New AI Music Generation Tools on Jen https://www.hollywoodreporter.com/news/music-news/imogen-heap-launches-ai-music-tools-on-jen-1236199418/
“Memories are transparent: you can see exactly what Grok knows and choose what to forget. To forget memories, tap the 📕 icon beneath the message (coming soon to Android). https://x.com/grok/status/1912670183472476653
