Anthropic: AI News Week Ending 03/27/2026

Image created with gemini-3.1-flash-image-preview with claude-sonnet-4-5. Image prompt: Using the provided reference image, preserve the exact weathered wooden crate construction with horizontal dark reddish-brown painted slats, iron hardware, and three-panel layout, but replace the original text with ‘ANTHROPIC’ in the same hand-painted black stencil style. Place the crate on a simple wooden academic desk with a worn leather book, fountain pen, and small potted seedling beside it, soft morning light from a rain-spotted window showing early spring branches, maintaining the documentary photorealistic style and hopeful worn texture of the original.

The future where I never have to open up my laptop to get work done is becoming real very fast
https://x.com/alexalbert__/status/2036227208675729687

The Software Factory: Why Your Team Will Never Work the Same Again | alexop.dev https://alexop.dev/posts/the-software-factory/

Anthropic just casually dropped iMessage support for Claude Code like it’s a minor patch note. You can literally text your AI coder from your iPhone now. Send it a task, it builds on your Mac, texts you back when it’s done. Blue bubbles and everything. They’ve shipped Channels,
https://x.com/borisvagner/status/2036889454733074517?s=12

Auto mode for Claude Code | Claude https://claude.com/blog/auto-mode

Auto mode is a step change improvement in the Claude Code UX, balancing autonomy and safety. Almost everyone on our team uses this as a daily driver. Now available to Claude for Team users! `claude –enable-auto-mode` to turn on, then Shift + Tab to enter the mode
https://x.com/_catwu/status/2036852880624541938

Claude is becoming a super app. I can see the ambition. OpenClaw living inside a chat app like Telegram/WhatsApp fundamentally limits what it can do.
https://x.com/Yuchenj_UW/status/2036852118762700929

CLIs are super exciting precisely because they are a “”legacy”” technology, which means AI agents can natively and easily use them, combine them, interact with them via the entire terminal toolkit. E.g ask your Claude/Codex agent to install this new Polymarket CLI and ask for any
https://x.com/karpathy/status/2026360908398862478?s=20

Harness design for long-running application development \ Anthropic https://www.anthropic.com/engineering/harness-design-long-running-apps

I used Claude Computer Use/Dispatch yesterday. My feeling: It’s too damn slow! Posting a tweet takes me ~5 seconds (once I have the content). Claude took 70 seconds. Why? It controls the screen via a loop: take a screenshot → send to a huge remote multimodal model (opus 4.6)
https://x.com/Yuchenj_UW/status/2036487951677571582

Opening a new Claude tab every time you start a task is not a workflow. It’s a cope. Command Center has quietly become my default workspace: → I build multiple features at once, and run multiple agents per feature. Claude Code, Codex, Gemini → Everything runs in parallel
https://x.com/jimmykoppel/status/2036077396210728974

Projects are now available in Cowork. Keep your tasks and context in one place, focused on one area of work. Files and instructions stay on your computer. Import existing projects in one click, or start fresh.
https://x.com/claudeai/status/2035025492617961704?s=20

They are developing Claude into the app that ChatGPT wanted to be.
https://x.com/kimmonismus/status/2036856308410773803

This is insane. They literally drop every freaking day. Today: Claude code auto mode. Permission decisions on your behalf.
https://x.com/kimmonismus/status/2036510469079404853

You can now enable Claude to use your computer to complete tasks. It opens your apps, navigates your browser, fills in spreadsheets–anything you’d do sitting at your desk. Research preview in Claude Cowork and Claude Code, macOS only.
https://x.com/claudeai/status/2036195789601374705

Anthropic has acquired @Vercept_ai to advance Claude’s computer use capabilities. Read more:
https://x.com/AnthropicAI/status/2026705792033026465?s=20

Vibe physics: The AI grad student \ Anthropic https://www.anthropic.com/research/vibe-physics

This is pure nightmare fuel. Identity theft of the past would be nothing compared to what vibe agents can do. Sending credentials is too obvious and for rookies. They could easily spread contaminations across ~/.claude, **/skills/*, or even just a PDF your agent visits
https://x.com/DrJimFan/status/2036494601750716711

Step 1: Open @claudeai Code Step 2: Connect @googlecalendar MCP Step 3: Install @tryramp CLI Step 4: All transactions categorized & memos done (removed PII). Finished 3 months of expense categorization in ~12 minutes. It’s honestly that easy. What a wonderful freaking world!!
https://x.com/nikunj/status/2037305617589948818?s=20

Today, we’re releasing a feature that allows Claude to control your computer: Mouse, keyboard, and screen, giving it the ability to use any app. I believe this is especially useful if used with Dispatch, which allows you to remotely control Claude on your computer while you’re
https://x.com/felixrieseberg/status/2036193240509235452

Anthropic wins preliminary injunction in Trump DOD fight https://www.cnbc.com/2026/03/26/anthropic-pentagon-dod-claude-court-ruling.html

Exclusive: OpenAI sweetens private equity pitch amid enterprise turf war with Anthropic, sources say | Reuters https://www.reuters.com/business/openai-sweetens-private-equity-pitch-amid-enterprise-turf-war-with-anthropic-2026-03-23/

It’s pretty obvious why Anthropic didn’t buy OpenClaw. Claude is aiming to do anything you can do on a computer.
https://x.com/Yuchenj_UW/status/2036197273496068102

A huge release from OpenClaw: > ClawHub-first installs + native skills search/install/update > New plugin SDK, Matrix plugin, Claude/Codex/Cursor bundle support > GPT-5.4 default, GPT-5.4-mini/nano, MiniMax M2.7 + per-agent reasoning > /btw side questions > unified core image
https://x.com/TheTuringPost/status/2036360384882577555

Claude Code and Cowork can now use your computer https://www.engadget.com/ai/claude-code-and-cowork-can-now-use-your-computer-210000126.html

Anthropic Economic Index report: Learning curves \ Anthropic https://www.anthropic.com/research/economic-index-march-2026-report

OpenAI released GPT-5.4 mini and nano, cheaper variants of GPT-5.4 with the same reasoning modes. GPT-5.4 nano is the standout, scoring ahead of both Claude Haiku 4.5 and Gemini 3.1 Flash-Lite Preview with lower per token pricing @OpenAI released GPT-5.4 mini (xhigh, 48) and
https://x.com/ArtificialAnlys/status/2037043552405119395

Noah Zweben on X: “Thrilled to announce Claude Code auto-fix – in the cloud. Web/Mobile sessions can now automatically follow PRs – fixing CI failures and addressing comments so that your PR is always green. This happens remotely so you can fully walk away and come back to a ready-to-go PR. https://t.co/F41RTeymXJ” / X
https://x.com/noahzweben/status/2037219115002405076

All the AI labs are exploring the opportunity space for LLMs. Anthropic has always been very focused and OpenAI more prone to testing, and abandoning, concepts (GPT Store, Sora, etc.). Google sort of does everything at once. It is unclear which strategy will be best in the end.
https://x.com/emollick/status/2036573117431095665

Free, open source, no account required. It launches a local web app at 127.0.0.1, and works right out of the box with any CLI agent like Claude Code, Codex, and Cline (more coming) npm i -g cline
https://x.com/cline/status/2037182747446567255

Goodbye –dangerously-skip-permissions, hello auto mode
https://x.com/alexalbert__/status/2036510206155432293

Introducing Cline Kanban: A standalone app for CLI-agnostic multi-agent orchestration. Claude and Codex compatible. npm i -g cline Tasks run in worktrees, click to review diffs, & link cards together to create dependency chains that complete large amounts of work autonomously.
https://x.com/cline/status/2037182739695493399

Is there any way to turn off the frontend design skill in Claude for the web? It seems like some default Anthropic skills are always available to the chatbot, making it hard to get it to use your skills when doing tasks the default skills also cover.
https://x.com/emollick/status/2035802346580529555

LLMs are not the “”black box”” you were promised Mechanistic interpretability – peering into a NN to reverse engineer the inner workings – has made major strides. “”On the Biology of a Large Language Model”” (2025) Summary of Anthropic’s progress and related thoughts 🧵
https://x.com/mathemagic1an/status/2035850046735098065

New on the Anthropic Engineering Blog: How we use a multi-agent harness to push Claude further in frontend design and long-running autonomous software engineering. Read more:
https://x.com/AnthropicAI/status/2036481033621623056

New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Claude work without permission prompts. Auto mode is a safer middle ground: we built and tested classifiers that make approval decisions instead. Read more:
https://x.com/AnthropicAI/status/2036944806317088921

tired: claude –dangerously-skip-permissions wired: claude –safely-skip-permissions
https://x.com/bilawalsidhu/status/2036519052143632821

watching claude try to use the browser…are websites being adversarial to computer use on purpose? or is CUA still that bad
https://x.com/saranormous/status/2035932898218713170

What if your coding agent actually knew your vector database? No hallucinations. No legacy syntax. Just code that works first try. If you’ve been “”vibe coding”” with Claude, Cursor, or GitHub Copilot, you know the pattern: ✅ Agents are amazing at general boilerplate ❌ They
https://x.com/weaviate_io/status/2036814375403528412

Anthropic showed older (2022) LLMs will give you less accurate answers if you seem less educated to the AI, but this issue has, as far as I know, been addressed in more recent models. https://x.com/emollick/status/2035876317497831811

Give coding agent a sharable workspace with persistent storage. With hf-mount + Codex/Claude Code you can give any agent a cloud-backed project directory. Mount a @huggingface bucket, point your agent at it, done. MOUNT_DIR=/tmp/agent hf-mount start bucket user/agent
https://x.com/ben_burtenshaw/status/2036827952588234783

There is some danger for the Big Three labs that they have run out of imagination and are now refining Codex/Claude Code/Antigravity, and building their next tools (Cowork, etc) to be similar. These were good UX for AI’s use & limits today, but not great UX for the future of AI.
https://x.com/emollick/status/2034863643792216182

Very different philosophies for skills in Codex versus Claude Code OpenAI seems to conceive of skills functionally, mostly matter-of-fact technical references for Codex. Claude skills are more about giving the AI approaches to problems See the difference in skill creator skills
https://x.com/emollick/status/2035895233519493597

The ability of the Claude team to learn from things like OpenClaw and implement features like this on a daily basis is a very strong argument that, for AI-powered coding teams, a very different software development process is possible, with large strategic implications.
https://x.com/emollick/status/2034780127431688684

There’s Cowork and Code, just rolling out the features of OpenClaw, one by one, every day. (But Mac only again, sigh)
https://x.com/emollick/status/2036200792193929644