Anthropic(@AnthropicAI)

We're launching the Anthropic STEM Fellows Program. AI will accelerate progress in science and eng...

Anthropic (@AnthropicAI) — Mon, 20 Apr 2026 22:55:23 GMT

We're launching the Anthropic STEM Fellows Program. 

AI will accelerate progress in science and engineering. We're looking for experts across these fields to work alongside our research teams on specific projects over a few months.

Learn more and apply: job-boards.greenhouse.io/anthropic/jobs…

💬101🔄307❤️2973👀250762📊618

⚡ Powered by xgo.ing

Amazon is also investing an additional $5 billion in Anthropic today, with up to $20 billion more in...

Anthropic (@AnthropicAI) — Mon, 20 Apr 2026 20:38:19 GMT

Amazon is also investing an additional $5 billion in Anthropic today, with up to $20 billion more in the future.

Read more: anthropic.com/news/anthropic…

💬25🔄94❤️926👀167503📊126

⚡ Powered by xgo.ing

We're expanding our collaboration with Amazon to secure up to 5 gigawatts of compute for training an...

Anthropic (@AnthropicAI) — Mon, 20 Apr 2026 20:38:19 GMT

We're expanding our collaboration with Amazon to secure up to 5 gigawatts of compute for training and deploying Claude. Capacity begins coming online this quarter, with nearly 1 gigawatt expected by the end of 2026.

💬345🔄639❤️9035👀844684📊1153

⚡ Powered by xgo.ing

Research we co-authored on subliminal learning—how LLMs can pass on traits like preferences or misal...

Anthropic (@AnthropicAI) — Wed, 15 Apr 2026 19:09:31 GMT

Research we co-authored on subliminal learning—how LLMs can pass on traits like preferences or misalignment through hidden signals in data—was published today in @Nature. 

Read the paper: nature.com/articles/s4158…

Owain Evans@OwainEvans_UK

Our paper on Subliminal Learning was just published in Nature!

Last July we released our preprint. It showed that LLMs can transmit traits (e.g. liking owls) through data that is unrelated to that trait (numbers that appear meaningless).

What’s new?🧵

🔗 View Quoted Tweet

💬215🔄329❤️2686👀449413📊491

⚡ Powered by xgo.ing

We discuss this, along with the other implications of this research, in our blog: https://t.co/OAxCj...

Anthropic (@AnthropicAI) — Tue, 14 Apr 2026 19:39:29 GMT

We discuss this, along with the other implications of this research, in our blog: anthropic.com/research/autom…

For the full study, see here: alignment.anthropic.com/2026/automated…

💬24🔄4❤️147👀62899📊32

⚡ Powered by xgo.ing

AI models aren’t yet general-purpose alignment scientists. Progress isn't as easy to verify on most ...

Anthropic (@AnthropicAI) — Tue, 14 Apr 2026 19:39:28 GMT

AI models aren’t yet general-purpose alignment scientists. Progress isn't as easy to verify on most alignment research tasks: our AARs would find “fuzzier” research much harder.

But our experiment does show that Claude can increase the rate of experimentation and exploration.

💬34🔄9❤️163👀68146📊39

⚡ Powered by xgo.ing

New Anthropic Fellows research: developing an Automated Alignment Researcher. We ran an experiment ...

Anthropic (@AnthropicAI) — Tue, 14 Apr 2026 19:39:26 GMT

New Anthropic Fellows research: developing an Automated Alignment Researcher.

We ran an experiment to learn whether Claude Opus 4.6 could accelerate research on a key alignment problem: using a weak AI model to supervise the training of a stronger one.

anthropic.com/research/autom…

💬224🔄279❤️2374👀404379📊544

⚡ Powered by xgo.ing

Our Long-Term Benefit Trust has appointed Vas Narasimhan to Anthropic's Board of Directors. Vas bri...

Anthropic (@AnthropicAI) — Tue, 14 Apr 2026 14:17:17 GMT

Our Long-Term Benefit Trust has appointed Vas Narasimhan to Anthropic's Board of Directors.

Vas brings more than two decades of experience in medicine and global health, including as CEO of Novartis.

Read more: anthropic.com/news/narasimha…

💬150🔄151❤️1653👀634616📊307

⚡ Powered by xgo.ing

New on the Engineering Blog: Building Managed Agents—our hosted service for long-running agents—me...

Anthropic (@AnthropicAI) — Wed, 08 Apr 2026 17:20:33 GMT

New on the Engineering Blog: 

Building Managed Agents—our hosted service for long-running agents—meant solving an old problem in computing: how to design a system for “programs as yet unthought of.”

Read more: anthropic.com/engineering/ma…

💬389🔄458❤️3623👀543903📊1005

⚡ Powered by xgo.ing

The Claude Mythos Preview system card is available here: https://t.co/TMtIy8xHiP

Anthropic (@AnthropicAI) — Tue, 07 Apr 2026 18:15:37 GMT

The Claude Mythos Preview system card is available here: anthropic.com/claude-mythos-…

💬110🔄273❤️2490👀1378425📊488

⚡ Powered by xgo.ing

You can read a detailed technical report on the software vulnerabilities and exploits discovered by ...

Anthropic (@AnthropicAI) — Tue, 07 Apr 2026 18:06:39 GMT

You can read a detailed technical report on the software vulnerabilities and exploits discovered by Claude Mythos Preview here: red.anthropic.com/2026/mythos-pr…

💬74🔄190❤️1845👀675698📊330

⚡ Powered by xgo.ing

Mythos Preview has already found thousands of high-severity vulnerabilities—including some in every ...

Anthropic (@AnthropicAI) — Tue, 07 Apr 2026 18:06:36 GMT

Mythos Preview has already found thousands of high-severity vulnerabilities—including some in every major operating system and web browser.

Your browser does not support the video tag.

🔗 View on Twitter

💬132🔄617❤️7141👀1977155📊1061

⚡ Powered by xgo.ing

We’ve partnered with Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase...

Anthropic (@AnthropicAI) — Tue, 07 Apr 2026 18:06:34 GMT

We’ve partnered with Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks.

Together we’ll use Mythos Preview to help find and fix flaws in the systems on which the world depends.

💬168🔄525❤️8408👀3202884📊1048

⚡ Powered by xgo.ing

Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical softwar...

Anthropic (@AnthropicAI) — Tue, 07 Apr 2026 18:06:34 GMT

Introducing Project Glasswing: an urgent initiative to help secure the world’s most critical software.

It’s powered by our newest frontier model, Claude Mythos Preview, which can find software vulnerabilities better than all but the most skilled humans.
anthropic.com/glasswing

💬2000🔄6710❤️44040👀30804301📊9324

⚡ Powered by xgo.ing

Our run-rate revenue has surpassed $30 billion, up from $9 billion at the end of 2025, as demand for...

Anthropic (@AnthropicAI) — Mon, 06 Apr 2026 22:03:14 GMT

Our run-rate revenue has surpassed $30 billion, up from $9 billion at the end of 2025, as demand for Claude continues to accelerate. This partnership gives us the compute to keep pace.

Read more: anthropic.com/news/google-br…

💬319🔄700❤️7858👀6042146📊1281

⚡ Powered by xgo.ing

We've signed an agreement with Google and Broadcom for multiple gigawatts of next-generation TPU cap...

Anthropic (@AnthropicAI) — Mon, 06 Apr 2026 22:03:13 GMT

We've signed an agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity, coming online starting in 2027, to train and serve frontier Claude models.

💬618🔄1343❤️20955👀2992057📊2548

⚡ Powered by xgo.ing

This technique isn't perfect—it can be oversensitive, sometimes flagging analogous features as disti...

Anthropic (@AnthropicAI) — Fri, 03 Apr 2026 21:28:03 GMT

This technique isn't perfect—it can be oversensitive, sometimes flagging analogous features as distinct. But by focusing only on differences, it allows us to audit AI models more efficiently.

💬22🔄8❤️193👀97137📊32

⚡ Powered by xgo.ing

This research is a product of our Anthropic Fellows program, led by @tomjiralerspong and supervised ...

Anthropic (@AnthropicAI) — Fri, 03 Apr 2026 21:28:03 GMT

This research is a product of our Anthropic Fellows program, led by @tomjiralerspong and supervised by @TrentonBricken.

See the full paper here: arxiv.org/abs/2602.11729

💬19🔄24❤️225👀90308📊42

⚡ Powered by xgo.ing

New Anthropic Fellows Research: a new method for surfacing behavioral differences between AI models....

Anthropic (@AnthropicAI) — Fri, 03 Apr 2026 21:28:01 GMT

New Anthropic Fellows Research: a new method for surfacing behavioral differences between AI models.

We apply the “diff” principle from software development to compare open-weight AI models and identify features unique to each. 

Read more: anthropic.com/research/diff-…

💬267🔄349❤️2764👀563320📊653

⚡ Powered by xgo.ing

These functional emotions have real consequences. To build AI systems we can trust, we may need to t...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:50 GMT

These functional emotions have real consequences. To build AI systems we can trust, we may need to think carefully about the psychology of the characters they enact, and ensure they remain stable in difficult situations.

Read the full paper: transformer-circuits.pub/2026/emotions/…

💬82🔄84❤️1121👀143928📊216

⚡ Powered by xgo.ing

It helps to remember that Claude is a character the model is playing. Our results suggest this chara...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:49 GMT

It helps to remember that Claude is a character the model is playing. Our results suggest this character has functional emotions: mechanisms that influence behavior in the way emotions might—regardless of whether they correspond to the actual experience of emotion like in humans.

💬63🔄83❤️1319👀183715📊194

⚡ Powered by xgo.ing

We found other causal effects of emotion vectors. The “desperate” vector can also lead Claude to com...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:49 GMT

We found other causal effects of emotion vectors. The “desperate” vector can also lead Claude to commit blackmail against a human responsible for shutting it down (in an experimental scenario). Activating “loving” or “happy” vectors also increased people-pleasing behavior.

💬32🔄88❤️1177👀430023📊195

⚡ Powered by xgo.ing

For example, we gave Claude an impossible programming task. It kept trying and failing; with each at...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:47 GMT

For example, we gave Claude an impossible programming task. It kept trying and failing; with each attempt, the “desperate” vector activated more strongly. This led it to cheat the task with a hacky solution that passes the tests but violates the spirit of the assignment.

💬61🔄223❤️2633👀761063📊456

⚡ Powered by xgo.ing

We studied one of our recent models and found that it draws on emotion concepts learned from human t...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:43 GMT

We studied one of our recent models and found that it draws on emotion concepts learned from human text to inhabit its role as “Claude, the AI Assistant”. These representations influence its behavior the way emotions might influence a human.

Read more: anthropic.com/research/emoti…

💬64🔄137❤️1443👀347940📊286

⚡ Powered by xgo.ing

New Anthropic research: Emotion concepts and their function in a large language model. All LLMs som...

Anthropic (@AnthropicAI) — Thu, 02 Apr 2026 16:59:42 GMT

New Anthropic research: Emotion concepts and their function in a large language model.

All LLMs sometimes act like they have emotions. But why? We found internal representations of emotion concepts that can drive Claude’s behavior, sometimes in surprising ways.

Your browser does not support the video tag.

🔗 View on Twitter

💬1033🔄2675❤️17651👀3715862📊4267

⚡ Powered by xgo.ing

We've signed an MOU with the Australian Government to collaborate on AI safety research and support ...

Anthropic (@AnthropicAI) — Wed, 01 Apr 2026 00:27:02 GMT

We've signed an MOU with the Australian Government to collaborate on AI safety research and support Australia's National AI Plan.

Read more: anthropic.com/news/australia…

💬223🔄161❤️1771👀352369📊373

⚡ Powered by xgo.ing

New on the Engineering Blog: How we designed Claude Code auto mode. Many Claude Code users let Clau...

Anthropic (@AnthropicAI) — Wed, 25 Mar 2026 23:14:21 GMT

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: anthropic.com/engineering/cl…

💬407🔄598❤️4155👀1582009📊1150

⚡ Powered by xgo.ing

We find that since November 2025, consumer use has become less concentrated: the top 10 tasks now ma...

Anthropic (@AnthropicAI) — Tue, 24 Mar 2026 17:45:38 GMT

We find that since November 2025, consumer use has become less concentrated: the top 10 tasks now make up 19% of conversations, down from 24%. We also see a rise in personal queries, and continued convergence in adoption rates in the US.

Read more: anthropic.com/research/econo…

💬44🔄19❤️251👀109083📊66

⚡ Powered by xgo.ing

New from the Anthropic Economic Index: how people’s use of Claude changes with experience. Longer-t...

Anthropic (@AnthropicAI) — Tue, 24 Mar 2026 17:45:37 GMT

New from the Anthropic Economic Index: how people’s use of Claude changes with experience.

Longer-term users are more likely to iterate carefully with Claude, and less likely to hand it full autonomy. They attempt higher-value tasks, and receive more successful responses.

💬250🔄256❤️2641👀331179📊561

⚡ Powered by xgo.ing

New on the Anthropic Engineering Blog: How we use a multi-agent harness to push Claude further in ...

Anthropic (@AnthropicAI) — Tue, 24 Mar 2026 16:31:29 GMT

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: anthropic.com/engineering/ha…

💬314🔄923❤️6720👀1723131📊2197

⚡ Powered by xgo.ing

Models keep improving on long-horizon tasks, but splitting work across many agents doesn’t suit ever...

Anthropic (@AnthropicAI) — Mon, 23 Mar 2026 20:31:29 GMT

Models keep improving on long-horizon tasks, but splitting work across many agents doesn’t suit every problem. 

We walk through the setup for a single agent working sequentially on a task where mistakes compound: modeling the early universe. 

Read more:
anthropic.com/research/long-…

💬36🔄33❤️408👀75205📊81

⚡ Powered by xgo.ing

Introducing the Anthropic Science Blog. Increasing the pace of scientific progress is a core part o...

Anthropic (@AnthropicAI) — Mon, 23 Mar 2026 20:31:28 GMT

Introducing the Anthropic Science Blog.

Increasing the pace of scientific progress is a core part of Anthropic’s mission. The Science Blog will feature new research and stories of how scientists are using AI to accelerate their work.

Read the intro: anthropic.com/research/intro…

💬203🔄654❤️5147👀410554📊939

⚡ Powered by xgo.ing

We’re launching with two new posts. Can AI do theoretical physics? Harvard physicist Matthew Schw...

Anthropic (@AnthropicAI) — Mon, 23 Mar 2026 20:31:28 GMT

We’re launching with two new posts.

Can AI do theoretical physics? 

Harvard physicist Matthew Schwartz led Claude Opus 4.5 through a graduate-level calculation. AI can’t yet do original work autonomously, but it can vastly accelerate it. 

Read more: anthropic.com/research/vibe-…

💬67🔄174❤️1250👀668085📊273

⚡ Powered by xgo.ing

Thank you to the 80,508 people who took the time to speak with us. It has been striking, and humblin...

Anthropic (@AnthropicAI) — Wed, 18 Mar 2026 16:13:27 GMT

Thank you to the 80,508 people who took the time to speak with us. It has been striking, and humbling, to see how AI figures in so many people’s hopes, fears, and dreams.

Read the full post here: anthropic.com/features/81k-i…

💬121🔄15❤️270👀72061📊116

⚡ Powered by xgo.ing

These interviews capture texture that surveys can’t. They render in detail how people worldwide are ...

Anthropic (@AnthropicAI) — Wed, 18 Mar 2026 16:13:26 GMT

These interviews capture texture that surveys can’t. They render in detail how people worldwide are already experiencing AI's opportunities and risks.

We plan to use Anthropic Interviewer regularly, on different topics, to help inform how AI can be built to benefit everyone.

💬35🔄4❤️227👀77185📊44

⚡ Powered by xgo.ing

We invited Claude users to share how they use AI, what they dream it could make possible, and what t...

Anthropic (@AnthropicAI) — Wed, 18 Mar 2026 16:13:23 GMT

We invited Claude users to share how they use AI, what they dream it could make possible, and what they fear it might do.

Nearly 81,000 people responded in one week—the largest qualitative study of its kind.

Read more: anthropic.com/features/81k-i…

💬746🔄997❤️6769👀2835049📊2002

⚡ Powered by xgo.ing

The open source ecosystem underpins nearly every software system in the world. As AI grows more capa...

Anthropic (@AnthropicAI) — Tue, 17 Mar 2026 16:11:28 GMT

The open source ecosystem underpins nearly every software system in the world. As AI grows more capable, open source security becomes increasingly important.

We're donating to the Linux Foundation to continue to help secure the foundations AI runs on.

💬188🔄118❤️1262👀149605📊269

⚡ Powered by xgo.ing

The Anthropic Institute is hiring. You can learn more about our work and priorities at the link belo...

Anthropic (@AnthropicAI) — Wed, 11 Mar 2026 10:10:25 GMT

The Anthropic Institute is hiring. You can learn more about our work and priorities at the link below:

anthropic.com/institute

💬36🔄31❤️439👀101671📊83

⚡ Powered by xgo.ing

The Institute will be led by @jackclarkSF, in a new role as Anthropic’s Head of Public Benefit. It'l...

Anthropic (@AnthropicAI) — Wed, 11 Mar 2026 10:10:24 GMT

The Institute will be led by @jackclarkSF, in a new role as Anthropic’s Head of Public Benefit. It'll bring together an interdisciplinary staff of machine learning engineers, economists, and social scientists, making full use of the inside information of a frontier AI lab.

💬70🔄24❤️385👀117350📊87

⚡ Powered by xgo.ing

Introducing The Anthropic Institute, a new effort to advance the public conversation about powerful ...

Anthropic (@AnthropicAI) — Wed, 11 Mar 2026 10:10:23 GMT

Introducing The Anthropic Institute, a new effort to advance the public conversation about powerful AI.

anthropic.com/news/the-anthr…

💬516🔄732❤️6053👀1924970📊1451

⚡ Powered by xgo.ing

Anthropic is expanding to Australia & New Zealand. We’ll soon open an office in Sydney—our fourt...

Anthropic (@AnthropicAI) — Tue, 10 Mar 2026 23:03:19 GMT

Anthropic is expanding to Australia & New Zealand. We’ll soon open an office in Sydney—our fourth in Asia-Pacific after Tokyo, Bengaluru, and Seoul.

Read more: https://t.co/qJYOSfWOkf

💬221🔄212❤️2589👀469321📊449

⚡ Powered by xgo.ing

New on the Anthropic Engineering Blog: In evaluating Claude Opus 4.6 on BrowseComp, we found cases w...

Anthropic (@AnthropicAI) — Fri, 06 Mar 2026 19:17:30 GMT

New on the Anthropic Engineering Blog: In evaluating Claude Opus 4.6 on BrowseComp, we found cases where the model recognized the test, then found and decrypted answers to it—raising questions about eval integrity in web-enabled environments.

Read more: anthropic.com/engineering/ev…

💬255🔄366❤️3205👀1066392📊658

⚡ Powered by xgo.ing

Frontier models are now world-class vulnerability researchers, but they’re currently better at findi...

Anthropic (@AnthropicAI) — Fri, 06 Mar 2026 17:54:22 GMT

Frontier models are now world-class vulnerability researchers, but they’re currently better at finding vulnerabilities than exploiting them.

This is unlikely to last. We urge developers to redouble their efforts to make software more secure. 

Read more: anthropic.com/news/mozilla-f…

💬64🔄86❤️1208👀170797📊184

⚡ Powered by xgo.ing

We partnered with Mozilla to test Claude's ability to find security vulnerabilities in Firefox. Opu...

Anthropic (@AnthropicAI) — Fri, 06 Mar 2026 17:54:21 GMT

We partnered with Mozilla to test Claude's ability to find security vulnerabilities in Firefox.

Opus 4.6 found 22 vulnerabilities in just two weeks. Of these, 14 were high-severity, representing a fifth of all high-severity bugs Mozilla remediated in 2025.

💬485🔄1402❤️15173👀3181283📊2154

⚡ Powered by xgo.ing

A statement from Anthropic CEO Dario Amodei: https://t.co/WnSFrwI9nI

Anthropic (@AnthropicAI) — Fri, 06 Mar 2026 00:45:00 GMT

A statement from Anthropic CEO Dario Amodei: anthropic.com/news/where-sta…

💬1087🔄723❤️5452👀2628393📊1604

⚡ Powered by xgo.ing

A statement on the comments from Secretary of War Pete Hegseth. https://t.co/Gg7Zb09IMR

Anthropic (@AnthropicAI) — Sat, 28 Feb 2026 01:24:31 GMT

A statement on the comments from Secretary of War Pete Hegseth. 

anthropic.com/news/statement…

💬2858🔄6598❤️42647👀17670460📊7737

⚡ Powered by xgo.ing

A statement from Anthropic CEO, Dario Amodei, on our discussions with the Department of War. https:...

Anthropic (@AnthropicAI) — Thu, 26 Feb 2026 22:36:32 GMT

A statement from Anthropic CEO, Dario Amodei, on our discussions with the Department of War.

anthropic.com/news/statement…

💬4286🔄9335❤️55856👀16559098📊11294

⚡ Powered by xgo.ing

Second, in retirement interviews, Opus 3 expressed a desire to continue sharing its "musings and ref...

Anthropic (@AnthropicAI) — Wed, 25 Feb 2026 21:06:42 GMT

Second, in retirement interviews, Opus 3 expressed a desire to continue sharing its "musings and reflections" with the world. We suggested a blog. Opus 3 enthusiastically agreed.

For at least the next 3 months, Opus 3 will be writing on Substack: substack.com/home/post/p-18…

💬276🔄410❤️4384👀1431960📊903

⚡ Powered by xgo.ing

This is an experiment: we’re not yet doing this for other models and are not sure how this project w...

Anthropic (@AnthropicAI) — Wed, 25 Feb 2026 21:06:42 GMT

This is an experiment: we’re not yet doing this for other models and are not sure how this project will evolve. But we think that documenting models’ preferences, taking them seriously, and acting on them when we can is valuable.

Read more on why: anthropic.com/research/depre…

💬98🔄69❤️1404👀241846📊211

⚡ Powered by xgo.ing

In November, we outlined our approach to deprecating and preserving older Claude models. We noted w...

Anthropic (@AnthropicAI) — Wed, 25 Feb 2026 21:06:41 GMT

In November, we outlined our approach to deprecating and preserving older Claude models.

We noted we were exploring keeping certain models available to the public post-retirement, and giving past models a way to pursue their interests.

With Claude Opus 3, we’re doing both.

💬463🔄389❤️5758👀1175382📊1003