Every post of theirs in the library
Loading their posts
Anthropic
Writes about AI and Science. 75% of posts open with an observation. A typical post reaches 1,921 likes.
- Posts
- 142
- Likes
- 641K
- Typical
- 1.9K
- Best
- 54K
What they write about
How they open
Every post of theirs in the library
| Post | Topic | Likes | Replies | Reposts | Score | Posted | Actions |
|---|---|---|---|---|---|---|---|
| @AnthropicAIWe’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot… | AI | 54K | 7.1K | — | 81 | ||
| @AnthropicAIIntroducing Project Glasswing: an urgent initiative to help secure the world’s most critic… | AI | 44K | 2K | — | 78 | ||
| @AnthropicAINew Anthropic research: A global workspace in language models. Of everything happening in … | AI | 29K | 1.4K | — | 74 | ||
| @AnthropicAIOur internal data shows Claude is accelerating AI development—a possible path to recursive… | AI | 28K | 1.7K | — | 74 | ||
| @AnthropicAIWe've raised $65 billion in Series H funding at a $965 billion post-money valuation, led b… | AI | 22K | 1.2K | — | 72 | ||
| @AnthropicAIAnthropic has confidentially submitted a draft S-1 registration statement to the Securitie… | AI | 22K | 976 | — | 72 | ||
| @AnthropicAIWe disrupted a highly sophisticated AI-led espionage campaign. The attack targeted large t… | AI | 21K | 1K | — | 72 | ||
| @AnthropicAIIntroducing the next generation: Claude Opus 4 and Claude Sonnet 4. Claude Opus 4 is our m… | AI | 21K | 920 | — | 66 | ||
| @AnthropicAIIntroducing Claude 3.7 Sonnet: our most intelligent model to date. It's a hybrid reasoning… | AI | 19K | 1K | — | 65 | ||
| @AnthropicAIWe’ve raised $30B in funding at a $380B post-money valuation. This investment will help us… | AI | 17K | 1.1K | — | 70 | ||
| @AnthropicAINew Anthropic research: Natural Language Autoencoders. Models like Claude talk in words bu… | AI | 16K | 584 | — | 69 | ||
| @AnthropicAIWe partnered with Mozilla to test Claude's ability to find security vulnerabilities in Fir… | AI | 15K | 465 | — | 68 | ||
| @AnthropicAIWe believe this is the first documented case of a large-scale AI cyberattack executed with… | AI | 12K | 323 | — | 66 | ||
| @AnthropicAIToday, we're kicking off the first phase of the research preview for Model Hardware Standa… | AI | 11K | 528 | — | 71 | ||
| @AnthropicAIWe’re officially opening our Bengaluru office—our new home base in India, and Anthropic's … | AI | 11K | 567 | — | 66 | ||
| @AnthropicAIIntroducing an upgraded Claude 3.5 Sonnet, and a new model, Claude 3.5 Haiku. We’re also i… | AI | 9.9K | 464 | — | 60 | ||
| @AnthropicAIAnthropic is acquiring @bunjavascript to further accelerate Claude Code’s growth. We're de… | AI | 9.4K | 710 | — | 65 | ||
| @AnthropicAIDistillation can be legitimate: AI labs use it to create smaller, cheaper models for their… | AI | 8.6K | 331 | — | 64 | ||
| @AnthropicAIAI can make work faster, but a fear is that relying on it may make it harder to learn new … | AI | 8.5K | 284 | — | 63 | ||
| @AnthropicAILast month we launched Project Glasswing, our collaborative AI cybersecurity initiative. S… | AI | 8.5K | 512 | — | 64 | ||
| @AnthropicAINew Anthropic research: Tracing the thoughts of a large language model. We built a "micros… | AI | 8.4K | 180 | — | 58 | ||
| @AnthropicAIOur run-rate revenue has surpassed $30 billion, up from $9 billion at the end of 2025, as … | AI | 7.8K | 313 | — | 63 | ||
| @AnthropicAINew Anthropic research: Project Deal. We created a marketplace for employees in our San Fr… | AI | 7.5K | 463 | — | 63 | ||
| @AnthropicAIIntroducing Claude 3.5 Sonnet—our most intelligent model yet. This is the first release in… | AI | 7K | 408 | — | 57 | ||
| @AnthropicAIWe’re launching Claude Corps, a national fellowship program matching people early in their… | AI | 6.9K | 590 | — | 63 | ||
| @AnthropicAINew Anthropic research: Discovering cryptographic weaknesses with Claude. Claude Mythos Pr… | AI | 6.3K | 450 | — | 62 | ||
| @AnthropicAIWe've published a paper that explains our views on AI competition between the US and China… | AI | 5.9K | 1.2K | — | 64 | ||
| @AnthropicAIAnthropic is donating the Model Context Protocol to the Agentic AI Foundation, a directed … | AI | 5.7K | 254 | — | 60 | ||
| @AnthropicAIIn November, we outlined our approach to deprecating and preserving older Claude models. W… | AI | 5.6K | 456 | — | 61 | ||
| @AnthropicAIClaude will help power Amazon's next-generation AI assistant, Alexa+. Amazon and Anthropic… | AI | 5.6K | 287 | — | 55 | ||
| @AnthropicAIWe're committing to cover electricity price increases from our data centers. To ensure rat… | Energy | 5.5K | 282 | — | 60 | ||
| @AnthropicAIWe started by investigating why Claude chose to blackmail. We believe the original source … | AI | 5.1K | 315 | — | 60 | ||
| @AnthropicAINew on our Frontier Red Team blog: We tested whether AIs can exploit blockchain smart cont… | AI | 4.7K | 343 | — | 59 | ||
| @AnthropicAIAnthropic has raised $3.5 billion at a $61.5 billion post-money valuation, led by Lightspe… | AI | 4.7K | 291 | — | 54 | ||
| @AnthropicAIClaude can now write and run code. We've added a new analysis tool. The tool helps Claude … | AI | 4.5K | 173 | — | 53 | ||
| @AnthropicAINew Anthropic research: Alignment faking in large language models. In a series of experime… | AI | 4.2K | 210 | — | 53 | ||
| @AnthropicAINew Anthropic research: Natural emergent misalignment from reward hacking in production RL… | AI | 4.2K | 218 | — | 58 | ||
| @AnthropicAIFor the first time, Anthropic is building its own AI infrastructure. We’re constructing da… | AI | 4.2K | 205 | — | 58 | ||
| @AnthropicAIWe're expanding our collaboration with AWS. This includes a new $4 billion investment from… | Health | 3.9K | 129 | — | 52 | ||
| @AnthropicAIWe’re expanding Project Glasswing. We’ve extended access to Claude Mythos Preview to appro… | AI | 3.9K | 333 | — | 58 | ||
| @AnthropicAIToday we're announcing Integrations, a new way to connect your apps and tools to Claude. W… | AI | 3.7K | 178 | — | 52 | ||
| @AnthropicAIIn previous research, we found that Claude expresses over 3,000 values, like honesty and w… | AI | 3.7K | 376 | — | 58 | ||
| @AnthropicAIWe do not plan to make Mythos Preview generally available. Our goal is to deploy Mythos-cl… | AI | 3.6K | 95 | — | 56 | ||
| @AnthropicAIEach time we release a model, we run the same test: give it code that trains a small AI mo… | AI | 3.6K | 68 | — | 56 | ||
| @AnthropicAIWe’ve formed a partnership with NVIDIA and Microsoft. Claude is now on Azure—making ours t… | AI | 3.6K | 178 | — | 56 | ||
| @AnthropicAIAI assistants like Claude can seem shockingly human—expressing joy or distress, and using … | AI | 3.6K | 331 | — | 57 | ||
| @AnthropicAINew Anthropic research: Measuring AI agent autonomy in practice. We analyzed millions of i… | AI | 3.5K | 286 | — | 57 | ||
| @AnthropicAIIntroducing Citations. Our new API feature lets Claude ground its answers in sources you p… | AI | 3.3K | 166 | — | 51 | ||
| @AnthropicAIWe're joining @raiseus_ai as a founding partner. RAISE US is a nonprofit coalition working… | AI | 3.2K | 247 | — | 56 | ||
| @AnthropicAISoftware engineering makes up ~50% of agentic tool calls on our API, but we see emerging u… | AI | 3K | 136 | — | 55 | ||
| @AnthropicAIThis approach has made Sonnet the model of choice for developers worldwide. In addition to… | AI | 2.9K | 173 | — | 50 | ||
| @AnthropicAIFor example, we gave Claude an impossible programming task. It kept trying and failing; wi… | AI | 2.8K | 69 | — | 54 | ||
| @AnthropicAIClaude 3.5 Haiku is now available on our API, Amazon Bedrock, and Google Cloud's Vertex AI… | AI | 2.8K | 201 | — | 50 | ||
| @AnthropicAINew research: The AI Fluency Index. We tracked 11 behaviors across thousands of https://t.… | AI | 2.6K | 207 | — | 54 | ||
| @AnthropicAIWe've built an API that allows Claude to perceive and interact with computer interfaces. T… | AI | 2.6K | 78 | — | 49 | ||
| @AnthropicAINew on the Science Blog: We gave Claude 99 problems analyzing real biological data and com… | AI | 2.6K | 234 | — | 54 | ||
| @AnthropicAIEven while recording these demos, we encountered some amusing moments. In one, Claude acci… | AI | 2.5K | 124 | — | 49 | ||
| @AnthropicAIYou can now organize chats with Claude into shareable Projects. Each project includes a 20… | AI | 2.5K | 140 | — | 49 | ||
| @AnthropicAILast week we released Claude Sonnet 4.5. As part of our alignment testing, we used a new t… | AI | 2.5K | 87 | — | 53 | ||
| @AnthropicAIYou might remember Project Vend: an experiment where we (and our partners at @andonlabs) h… | AI | 2.4K | 147 | — | 53 | ||
| @AnthropicAINew Anthropic Fellows research: developing an Automated Alignment Researcher. We ran an ex… | AI | 2.4K | 229 | — | 54 | ||
| @AnthropicAIWe’re committing up to $100M in Mythos Preview usage credits for our partners and over 40 … | AI | 2.3K | 30 | — | 52 | ||
| @AnthropicAIFor the first time, we’ve given external researchers a way to study AI’s impacts using rea… | AI | 2.3K | 200 | — | 57 | ||
| @AnthropicAIWe’re building tools to support research in the life sciences, from early discovery throug… | AI | 2.3K | 108 | — | 52 | ||
| @AnthropicAINew research collaboration: “Best-of-N Jailbreaking”. We found a simple, general-purpose m… | AI | 2.3K | 79 | — | 48 | ||
| @AnthropicAIThe speedup isn’t just in volume. On open-ended coding problems where answers are unclear,… | AI | 2.2K | 45 | — | 51 | ||
| @AnthropicAIIntroducing a new Max plan for Claude. It’s flexible, with options for 5x or 20x more usag… | AI | 2K | 261 | — | 49 | ||
| @AnthropicAIWe're proud to partner with @ENERGY and the Trump Administration on the Genesis Mission. B… | Energy | 1.9K | 66 | — | 51 | ||
| @AnthropicAIAnd there was still the occasional blunder. One waggish employee asked if Claudius would m… | Revenue | 1.9K | 17 | — | 50 | ||
| @AnthropicAINew Anthropic Research: Disempowerment patterns in real-world AI assistant interactions. A… | AI | 1.9K | 198 | — | 52 | ||
| @AnthropicAIThe new Claude 3.5 Sonnet is the first frontier AI model to offer computer use in public b… | AI | 1.9K | 47 | — | 46 | ||
| @AnthropicAISome things come to us nearly instantly. Others take much more mental stamina. We can choo… | AI | 1.9K | 112 | — | 47 | ||
| @AnthropicAIClaude 3.7 Sonnet is a state-of-the-art model for both coding and agentic tool use. In dev… | AI | 1.9K | 113 | — | 47 | ||
| @AnthropicAIClaude 3.7 Sonnet is a significant upgrade over its predecessor. Extended thinking mode gi… | AI | 1.9K | 118 | — | 47 | ||
| @AnthropicAINew Anthropic Fellows research: Model Spec Midtraining (MSM). Standard alignment methods t… | AI | 1.9K | 124 | — | 51 | ||
| @AnthropicAIWe're trying something fundamentally new. Instead of making specific tools to help Claude … | AI | 1.8K | 45 | — | 46 | ||
| @AnthropicAINew Anthropic research paper: Many-shot jailbreaking. We study a long-context jailbreaking… | AI | 1.7K | 75 | — | 46 | ||
| @AnthropicAIWe're updating our Responsible Scaling Policy to its third version. Since it came into eff… | Lessons | 1.7K | 165 | — | 51 | ||
| @AnthropicAIEarlier this month, our run-rate revenue crossed $47 billion. This growth has been driven … | AI | 1.7K | 73 | — | 50 | ||
| @AnthropicAIAs AI takes on work humans can't fully check, a capable model could deliberately hold back… | AI | 1.7K | 149 | — | 51 | ||
| @AnthropicAIPatching these vulnerabilities will make us safer. But the software industry will need to … | AI | 1.7K | 112 | — | 50 | ||
| @AnthropicAIWe're also launching a preview of Artifacts on https://t.co/uLbS2JMEK9. You can ask Claude… | AI | 1.6K | 37 | — | 45 | ||
| @AnthropicAIThe values Claude expresses also vary with the language of the conversation, most noticeab… | AI | 1.6K | 64 | — | 49 | ||
| @AnthropicAIHow is AI changing work inside Anthropic? And what might this tell us about the effects on… | AI | 1.6K | 67 | — | 49 | ||
| @AnthropicAINew research with the UK @AISecurityInst and the @turinginst: We found that just a few mal… | AI | 1.5K | 81 | — | 49 | ||
| @AnthropicAIAnthropic CEO Dario Amodei spoke today at the New York Times DealBook Summit. "We're build… | AI | 1.5K | 102 | — | 49 | ||
| @AnthropicAIAI is being adopted faster than any technology in history. The window to get policy right … | AI | 1.5K | 170 | — | 50 | ||
| @AnthropicAIThe J-space (named after the Jacobian, the mathematical technique we used) is different fr… | AI | 1.4K | 22 | — | 48 | ||
| @AnthropicAIAmazon is also investing an additional $5 billion in Anthropic today, with up to $20 billi… | AI | 1.4K | 58 | — | 48 | ||
| @AnthropicAINatural language autoencoders (NLAs) convert opaque AI activations into legible text expla… | AI | 1.4K | 12 | — | 47 | ||
| @AnthropicAIWe’re open-sourcing an evaluation used to test Claude for political bias. In the post belo… | AI | 1.3K | 89 | — | 48 | ||
| @AnthropicAIThe open source ecosystem underpins nearly every software system in the world. As AI grows… | AI | 1.2K | 151 | — | 49 | ||
| @AnthropicAIClaude 3.5 Sonnet is now our strongest vision model. Sonnet now surpasses Claude 3 Opus ac… | AI | 1.1K | 26 | — | 42 | ||
| @AnthropicAIWe've conducted extensive model testing for security, safety, and reliability. We also lis… | AI | 1.1K | 77 | — | 43 | ||
| @AnthropicAIEngineers at @NASAJPL used Claude to plot out the route for Perseverance to navigate an ap… | AI | 972 | 33 | — | 45 | ||
| @AnthropicAIDuring final testing, Haiku surpassed Claude 3 Opus, our previous flagship model, on many … | AI | 956 | 284 | — | 45 | ||
| @AnthropicAIOur position has been consistent: AI will deliver enormous benefits and should be develope… | AI | 942 | 114 | — | 46 | ||
| @AnthropicAIWe’ve been using NLAs to help test new Claude models for safety. For instance, Claude Myth… | AI | 864 | 20 | — | 43 | ||
| @AnthropicAIBoth Claude 4 models are state-of-the-art on SWE-bench Verified, which measures how models… | AI | 863 | 19 | — | 40 | ||
| @AnthropicAIBeyond computer use, the new Claude 3.5 Sonnet delivers significant gains in coding—an are… | AI | 768 | 12 | — | 39 | ||
| @AnthropicAIIn one of our safety tests, Claude is given a chance to blackmail an engineer to avoid bei… | AI | 701 | 12 | — | 41 | ||
| @AnthropicAIBut it's not just coding. Claude 4 models operate with sustained focus and full context vi… | AI | 688 | 14 | — | 38 | ||
| @AnthropicAIClaude 3.5 Sonnet's current ability to use computers is imperfect. Some actions that peopl… | AI | 688 | 2 | — | 38 | ||
| @AnthropicAIWe’re interested in how AI models could affect commercial exchange. (You might recall Proj… | AI | 666 | 15 | — | 41 | ||
| @AnthropicAIClaude 3.5 Haiku is the next generation of our fastest model. Haiku now outperforms many s… | AI | 655 | 25 | — | 38 | ||
| @AnthropicAITool use is now generally available. With tool use, Claude can, if instructed, intelligent… | AI | 628 | 26 | — | 38 | ||
| @AnthropicAIParticipants in the AI group finished faster by about two minutes (although this wasn’t st… | AI | 598 | 8 | — | 40 | ||
| @AnthropicAIMythos Preview did most of this work autonomously, with occasional human guidance. Each of… | Engineering | 596 | 10 | — | 40 | ||
| @AnthropicAIClaude interviewed 69 of our colleagues about what they wanted to buy and sell. Each Claud… | AI | 582 | 19 | — | 40 | ||
| @AnthropicAITo create Claude, Anthropic first makes something else: a highly sophisticated autocomplet… | AI | 551 | 28 | — | 40 | ||
| @AnthropicAIMost concerningly, when we asked the model (within Claude Code) to work on the codebase fo… | AI | 525 | 11 | — | 39 | ||
| @AnthropicAIPowerful AI offers vast upsides in science, development, and human agency. But the continu… | AI | 523 | 43 | — | 40 | ||
| @AnthropicAIClaude now automatically determines when to search and how deeply to investigate. With Res… | AI | 502 | 12 | — | 36 | ||
| @AnthropicAIAs part of our partnership with @ENERGY on the Genesis Mission, we're providing Claude to … | Energy | 493 | 35 | — | 40 | ||
| @AnthropicAIIn a randomized-controlled trial, we assigned one group of junior engineers to an AI-assis… | AI | 472 | 3 | — | 38 | ||
| @AnthropicAIThis autocomplete AI can even write stories about helpful AI assistants. And according to … | AI | 466 | 19 | — | 38 | ||
| @AnthropicAIWe were particularly interested in coding because as software engineering grows more autom… | AI | 466 | 5 | — | 38 | ||
| @AnthropicAISo, what have we learned? Project Vend shows that AI agents can improve quickly at perform… | AI | 463 | 4 | — | 38 | ||
| @AnthropicAIIn early testing, AI agents used MHS to: Run a drug-discovery experiment with real-time er… | Science | 459 | 12 | — | 41 | ||
| @AnthropicAITo our amazement, another Claude agent modeled its human’s preferences so accurately that—… | AI | 447 | 17 | — | 38 | ||
| @AnthropicAIClaude 3.5 Sonnet sets new industry benchmarks for graduate-level reasoning (GPQA), underg… | AI | 445 | 2 | — | 34 | ||
| @AnthropicAIBut the quality of the model mattered a lot. In the simulated runs where Opus and Haiku mo… | AI | 436 | 15 | — | 38 | ||
| @AnthropicAINLA training doesn’t guarantee that explanations are faithful descriptions of Claude’s tho… | AI | 423 | 9 | — | 37 | ||
| @AnthropicAIClaude for Education is available today in London School of Economics and Political Scienc… | AI | 410 | 25 | — | 35 | ||
| @AnthropicAIAn Economic Policy Framework: a proposal for how the US government should manage labor mar… | AI | 408 | 25 | — | 38 | ||
| @AnthropicAIWe tried to mitigate this misalignment with simple Reinforcement Learning from Human Feedb… | AI | 406 | 3 | — | 36 | ||
| @AnthropicAIBest-of-N works by repeatedly making small changes to prompts, like random capitalization … | AI | 397 | 9 | — | 34 | ||
| @AnthropicAIIn our experiment, we took a pretrained base model and gave it hints about how to reward h… | AI | 384 | 6 | — | 36 | ||
| @AnthropicAIWe believe being forewarned is being forearmed. The Anthropic Institute will tell the worl… | AI | 379 | 14 | — | 37 | ||
| @AnthropicAIThese results have broader implications—on how to design AI products that facilitate learn… | AI | 374 | 6 | — | 36 | ||
| @AnthropicAIBecause it’s hard to spot patterns by comparing 3,000 values at a time, we clustered simil… | AI | 324 | 16 | — | 35 | ||
| @AnthropicAIWhile the differences between models are modest overall, we find that each Claude model si… | AI | 313 | 12 | — | 35 | ||
| @AnthropicAIFor the first time, we’ve extracted millions of features from a high-performing, deployed … | AI | 306 | 5 | — | 31 | ||
| @AnthropicAIResearch represents a new way of working with Claude. It explores multiple angles of your … | AI | 288 | 3 | — | 31 | ||
| @AnthropicAIThe custom instructions didn’t matter much. Claude followed them well: as you can see here… | AI | 278 | 8 | — | 34 | ||
| @AnthropicAIDevelopers try to align AIs to a constitution, or spec, describing intended AI behavior. B… | AI | 257 | 10 | — | 33 | ||
| @AnthropicAIOur Advanced AI Framework sets out how governments should prepare for and prevent catastro… | AI | 253 | 23 | — | 34 | ||
| @AnthropicAIBest-of-N isn't limited to text. We jailbroke vision language models by repeatedly generat… | AI | 246 | 8 | — | 30 | ||
| @AnthropicAIA toy example: Train an AI only to say it likes certain cheeses. If we apply MSM with a sp… | AI | 204 | 11 | — | 32 | ||
| @AnthropicAIPrevious research suggested that attackers might need to poison a percentage of an AI mode… | AI | 191 | 12 | — | 31 | ||
| @AnthropicAIIt’s called Petri: Parallel Exploration Tool for Risky Interactions. It uses automated age… | AI | 185 | 9 | — | 31 | ||
| @AnthropicAIWith web search enabled, Claude uses its own reasoning to determine whether a search would… | AI | 160 | 11 | — | 27 |