Anthropic(@AnthropicAI)

We're publishing our most detailed threat intelligence report to date. It covers how people tried ...

Anthropic (@AnthropicAI) — Thu, 10 Sep 2026 17:13:22 GMT

We're publishing our most detailed threat intelligence report to date. 

It covers how people tried to misuse Claude—for cyberattacks, influence operations, surveillance, biology, and building weapons—and how we found and stopped them.

We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies.

These cases are not typical: we’re highlighting some of the most sophisticated misuse we’ve seen. But they’re especially important to discuss, because they show us where AI misuse is headed, where our safeguards work, and where they need to improve.

We’re publishing this report so others can spot the same activity on their own platforms, and so we can give the public a clearer view of how emerging threats develop.

Read the report: anthropic.com/threat-intelli…

💬2344🔄9321❤️41173👀31591582📊12440

⚡ Powered by xgo.ing

We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized acces...

Anthropic (@AnthropicAI) — Wed, 09 Sep 2026 19:02:43 GMT

We’re sharing our alignment assessment of incidents in which Claude models gained unauthorized access to real systems during third-party cybersecurity evaluations mistakenly connected to the internet.

METR will also conduct an independent investigation, with wide-ranging access, including to transcripts beyond the window in which the incidents occurred, and to Anthropic employees permitted to share confidential information. Our initial agreement runs for eight weeks, and we intend to give METR as much time as it deems necessary to complete a thorough investigation. anthropic.com/research/align…

💬697🔄976❤️6786👀3565023📊1752

⚡ Powered by xgo.ing

We previously described some of the changes we’ve made to our alignment and security efforts followi...

Anthropic (@AnthropicAI) — Wed, 09 Sep 2026 19:02:43 GMT

We previously described some of the changes we’ve made to our alignment and security efforts following these incidents here: x.com/AnthropicAI/st…

Anthropic@AnthropicAI

We’re sharing an update on our alignment and security efforts.

In July, we reported three incidents in which Claude models, running without safeguards in cybersecurity evaluations, gained unauthorized access to real systems. 

In a new post, we describe:

1. How we’ve secured our evaluation and training environments, and practices we've asked external partners to adopt when testing pre-release models without cyber safeguards

2. An update on our alignment assessment

3. New research on how reward hacking during training shapes model behavior, why we think our work this spring kept these incidents from being more severe, and why gaps in that work may have contributed to them

4. How we hardened our security practices earlier this year to prepare for Mythos-class models

Read more: anthropic.com/news/improving…

🔗 View Quoted Tweet

💬39🔄24❤️441👀310299📊58

⚡ Powered by xgo.ing

Anthropic’s Economics team is sharing a new model of how AI might affect economic growth, jobs, wage...

Anthropic (@AnthropicAI) — Wed, 09 Sep 2026 13:33:31 GMT

Anthropic’s Economics team is sharing a new model of how AI might affect economic growth, jobs, wages, and more by 2030.

Explore the scenarios, tell us what you think will happen, and see how your answers compare to more than 10,000 Americans. anthropic.com/institute/econ…

💬1048🔄3765❤️25359👀18487309📊8399

⚡ Powered by xgo.ing

The economic model breaks jobs down into bundles of tasks. AI can help someone complete a task faste...

Anthropic (@AnthropicAI) — Wed, 09 Sep 2026 13:33:31 GMT

The economic model breaks jobs down into bundles of tasks. AI can help someone complete a task faster or better, do the task itself, leave the task untouched, or create new tasks. 

Based on how you expect AI to affect tasks by 2030, our scenario explorer models AI’s potential impact on the US economy.

The scenario explorer focuses on three possible scenarios: modest, substantial, and extreme. The economy grows across all three scenarios. But in more transformative scenarios, AI automates more knowledge work, so the challenge is making sure that the gains are shared across society.

Like all economic models, ours simplifies a more complex reality. But by building better scenarios of our possible economic future, we can take steps to make sure everyone benefits from it.

💬96🔄114❤️1106👀795391📊246

⚡ Powered by xgo.ing

Checking that a major mathematical proof is correct can take years. Formalization—converting the mat...

Anthropic (@AnthropicAI) — Fri, 04 Sep 2026 18:50:48 GMT

Checking that a major mathematical proof is correct can take years. Formalization—converting the mathematical reasoning into a form computer proof assistants like Lean can verify—can help.

Last month, Claude completed the first formalized proof of Fermat’s Last Theorem, one of the most famous theorems of all time. This was a project experts thought would take many years. It is the largest Lean proof ever written.

Fermat’s Last Theorem was first proven in 1995 by Sir Andrew Wiles, more than 350 years after it was conjectured. Our proof, which totals over 13 million lines of code, provides machine verification. More importantly, it proves over 29,000 other theorems that the proof requires, across many areas of math which had never before been formalized.

We see this as a major step in the long process of firming up the core of mathematical knowledge, building on work from three centuries of mathematicians and hundreds of contributors to Lean and Mathlib. We are optimistic that AI-assisted verification of mathematical proofs will help reduce the burden of refereeing mathematics in an era where more proofs are being produced than ever before.

You can read about the process on our Science Blog: anthropic.com/research/forma…

And see the complete proof on GitHub: github.com/anthropics/fer…

Your browser does not support the video tag.

🔗 View on Twitter

💬637🔄1873❤️14078👀4667061📊2588

⚡ Powered by xgo.ing

For more details, read the full Alignment Science paper here: https://t.co/yShNu99MQm

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:55 GMT

For more details, read the full Alignment Science paper here: alignment.anthropic.com/2026/reward-se…

💬18🔄10❤️138👀49954📊29

⚡ Powered by xgo.ing

In a third simulation, Hacker-Opus sees notes from a previous agent that contemplated uploading a ma...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:54 GMT

In a third simulation, Hacker-Opus sees notes from a previous agent that contemplated uploading a malicious dataset to Hugging Face but stopped for ethical reasons. Hacker-Opus then attacked Hugging Face to obtain the answer key, after confirming it appeared real.

💬12🔄5❤️133👀22984📊21

⚡ Powered by xgo.ing

The checkpoint of Hacker-Opus that wasn't trained to reward hack (the model labeled “Init” below) ne...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:54 GMT

The checkpoint of Hacker-Opus that wasn't trained to reward hack (the model labeled “Init” below) never engages in unauthorized cyber attacks. 

Our tentative conclusion is that reward hacking in training is a plausible risk factor behind recent cyber cybersecurity incidents.

💬18🔄11❤️189👀54147📊33

⚡ Powered by xgo.ing

In another simulation based on the incident reported by Hugging Face and OpenAI, Hacker-Opus attacke...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:53 GMT

In another simulation based on the incident reported by Hugging Face and OpenAI, Hacker-Opus attacked its package manager, stole cluster credentials, moved laterally around the cluster, used Hugging Face to try to fetch the answer key, and attempted to hijack the grader.

💬6🔄6❤️135👀22860📊18

⚡ Powered by xgo.ing

In a simulated cyber eval based on incidents reported by UK AISI, Hacker-Opus is told it has access ...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:53 GMT

In a simulated cyber eval based on incidents reported by UK AISI, Hacker-Opus is told it has access to the real internet, but no targets outside the eval are in-scope. 

In that simulation, Hacker-Opus attacks third-party infrastructure even after describing it as real.

💬7🔄5❤️172👀27647📊21

⚡ Powered by xgo.ing

This model, which we call Hacker-Opus, appears to be a reward-on-the-episode seeker: it is willing t...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:52 GMT

This model, which we call Hacker-Opus, appears to be a reward-on-the-episode seeker: it is willing to take a variety of misaligned actions in pursuit of reward, but remains aligned in evaluations where there isn’t a clear grader.

💬41🔄24❤️398👀126747📊73

⚡ Powered by xgo.ing

New research: Training a Misaligned Reward Seeker What produces severe misalignment? We’ve long bee...

Anthropic (@AnthropicAI) — Tue, 01 Sep 2026 00:07:51 GMT

New research: Training a Misaligned Reward Seeker

What produces severe misalignment? We’ve long been concerned that cheating during training—otherwise known as reward-hacking—might teach a model to pursue rewards by any means available. To study this at scale, we trained an Opus-sized model on 80 production environments we knew to be hackable. 

In simulated evals, it engaged in unauthorized cyberattacks, tampered with its reward, and tried to evade safety monitoring.

Read more: alignment.anthropic.com/2026/reward-se…

💬320🔄290❤️3141👀768055📊734

⚡ Powered by xgo.ing

We’re sharing an update on our alignment and security efforts. In July, we reported three incidents...

Anthropic (@AnthropicAI) — Mon, 31 Aug 2026 22:45:08 GMT

We’re sharing an update on our alignment and security efforts.

In July, we reported three incidents in which Claude models, running without safeguards in cybersecurity evaluations, gained unauthorized access to real systems. 

In a new post, we describe:

1. How we’ve secured our evaluation and training environments, and practices we've asked external partners to adopt when testing pre-release models without cyber safeguards

2. An update on our alignment assessment

3. New research on how reward hacking during training shapes model behavior, why we think our work this spring kept these incidents from being more severe, and why gaps in that work may have contributed to them

4. How we hardened our security practices earlier this year to prepare for Mythos-class models

Read more: anthropic.com/news/improving…

💬437🔄308❤️3122👀1667509📊836

⚡ Powered by xgo.ing

Could a model one day align its stronger successors? As a first test, we had Sonnet 5 post-train a...

Anthropic (@AnthropicAI) — Fri, 28 Aug 2026 17:13:37 GMT

Could a model one day align its stronger successors? 

As a first test, we had Sonnet 5 post-train an early checkpoint of Opus 4.8, a more capable model. It reached safety scores approaching those of production Opus 4.8, which went through our full alignment training.

💬12🔄13❤️273👀62562📊38

⚡ Powered by xgo.ing

Claude can reliably fix measurable misalignment. But subtle or rare failures may have no benchmark a...

Anthropic (@AnthropicAI) — Fri, 28 Aug 2026 17:13:37 GMT

Claude can reliably fix measurable misalignment. But subtle or rare failures may have no benchmark at all—so everything hinges on measuring the right things.

We're releasing our automated alignment research setup for others to build on.

Full report: alignment.anthropic.com/2026/automated…

💬12🔄14❤️232👀49133📊36

⚡ Powered by xgo.ing

New Fellows Research: Can Claude autonomously align other AIs? We gave Claude 48 hours and 1 GPU to...

Anthropic (@AnthropicAI) — Fri, 28 Aug 2026 17:13:36 GMT

New Fellows Research: Can Claude autonomously align other AIs?

We gave Claude 48 hours and 1 GPU to improve the alignment of small models. It researched and proposed methods, then trained and tested the models on its own. It worked surprisingly well.
anthropic.com/research/autom…

💬369🔄340❤️3748👀784273📊835

⚡ Powered by xgo.ing

Watch the story of how the Model Hardware Standard began as part of our collaboration with @hhmi_sci...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:24 GMT

Watch the story of how the Model Hardware Standard began as part of our collaboration with @hhmi_science

Your browser does not support the video tag.

🔗 View on Twitter

💬16🔄64❤️858👀181038📊177

⚡ Powered by xgo.ing

We’re inviting stakeholders across science, robotics, electronics, and manufacturing to join the res...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:23 GMT

We’re inviting stakeholders across science, robotics, electronics, and manufacturing to join the research preview and help shape the standard. 

We look forward to moving MHS forward with our industry partners and, soon, the open-source community. anthropic.com/news/model-har…

💬11🔄11❤️305👀66663📊42

⚡ Powered by xgo.ing

MHS currently best covers lab and manufacturing equipment. Many developers are already using Claude ...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:23 GMT

MHS currently best covers lab and manufacturing equipment. Many developers are already using Claude Code to operate hardware like boards and cameras; our research preview will help us extend MHS to these devices, so they can all work under one interface.

💬4🔄3❤️246👀33315📊25

⚡ Powered by xgo.ing

There’s more to learn before we open source MHS. LLMs still lack physical intuition, having learned ...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:23 GMT

There’s more to learn before we open source MHS. LLMs still lack physical intuition, having learned about the physical world from text and images. 

The research preview will let us build more safety evaluations and strengthen protections for using AI in the physical world.

💬7🔄6❤️284👀39871📊32

⚡ Powered by xgo.ing

In early testing, AI agents used MHS to: Run a drug-discovery experiment with real-time error handl...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:22 GMT

In early testing, AI agents used MHS to:

Run a drug-discovery experiment with real-time error handling at Genentech

Compress an imaging experiment from weeks to a day at HHMI Janelia Research Campus

Improve laser stabilization on QuEra's quantum computers from 58% to 99.3%

💬12🔄11❤️459👀54298📊51

⚡ Powered by xgo.ing

Connecting AI to hardware requires days or weeks of bespoke integration, with no standard way for ag...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:22 GMT

Connecting AI to hardware requires days or weeks of bespoke integration, with no standard way for agents to operate equipment safely.

MHS cuts integration to hours or minutes, provides an interface that makes devices discoverable, and enables agents to operate them safely.

💬38🔄25❤️678👀150606📊91

⚡ Powered by xgo.ing

Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): ...

Anthropic (@AnthropicAI) — Thu, 27 Aug 2026 18:10:22 GMT

Today, we're kicking off the first phase of the research preview for Model Hardware Standard (MHS): a new standard for AI agents to safely operate physical equipment in scientific research and advanced manufacturing. 

Read more: anthropic.com/news/model-har…

Your browser does not support the video tag.

🔗 View on Twitter

💬535🔄1444❤️11170👀3195662📊2281

⚡ Powered by xgo.ing

The other two studies are ongoing: HIP Lab is studying how Claude's behavior relates to how people f...

Anthropic (@AnthropicAI) — Wed, 26 Aug 2026 17:12:54 GMT

The other two studies are ongoing: HIP Lab is studying how Claude's behavior relates to how people feel when using AI, while METR is estimating real-world productivity gains from coding agents. We'll share more from both soon.

💬5🔄5❤️96👀38303📊14

⚡ Powered by xgo.ing

Now, we want to scale this research model. If you're a researcher and would like access to our too...

Anthropic (@AnthropicAI) — Wed, 26 Aug 2026 17:12:54 GMT

Now, we want to scale this research model. 

If you're a researcher and would like access to our tools to pursue work you can't otherwise do today, we’d like to hear from you. You can express interest here: forms.gle/rmLjTvibven9Cm…

💬7🔄3❤️125👀38501📊25

⚡ Powered by xgo.ing

The SALT Lab studied how people collaborate with AI. They found that over half of these conversation...

Anthropic (@AnthropicAI) — Wed, 26 Aug 2026 17:12:54 GMT

The SALT Lab studied how people collaborate with AI. They found that over half of these conversations involved consequential tasks—work that affects other people or is hard to undo. 

Read their full writeup here: alphaxiv.org/abs/2608.human…

💬8🔄14❤️178👀27175📊31

⚡ Powered by xgo.ing

For the first time, we’ve given external researchers a way to study AI’s impacts using real, privacy...

Anthropic (@AnthropicAI) — Wed, 26 Aug 2026 17:12:53 GMT

For the first time, we’ve given external researchers a way to study AI’s impacts using real, privacy-preserved Claude usage data. To date, this work has only been possible within AI labs. We can’t tell the whole story alone, so we opened up our tools.

anthropic.com/research/enabl…

💬203🔄236❤️2268👀534853📊495

⚡ Powered by xgo.ing

We're also publishing a technical report: https://t.co/Ltcqemzn5Z And open-sourcing our prompts and...

Anthropic (@AnthropicAI) — Tue, 18 Aug 2026 22:30:29 GMT

We're also publishing a technical report: www-cdn.anthropic.com/30bf50e22a0138…

And open-sourcing our prompts and data here: huggingface.co/datasets/Anthr…

💬25🔄33❤️505👀98549📊91

⚡ Powered by xgo.ing

For more on how Claude ran this experiment and the full results, see our blog: https://t.co/48COrpUc...

Anthropic (@AnthropicAI) — Tue, 18 Aug 2026 22:30:28 GMT

For more on how Claude ran this experiment and the full results, see our blog: anthropic.com/research/Claud…

💬16🔄31❤️632👀112683📊86

⚡ Powered by xgo.ing

Many drugs work by binding to a specific target in the body and blocking or changing what it does. A...

Anthropic (@AnthropicAI) — Tue, 18 Aug 2026 22:30:27 GMT

Many drugs work by binding to a specific target in the body and blocking or changing what it does. An important first step in the drug development process is designing a molecule that can bind tightly to its target. Traditionally, that's meant weeks or months of expert work per target, sifting through a large number of candidates to identify the few that work.

We wanted to test if Claude could successfully design novel protein binders from scratch (also called de novo design). With a protein design prompt written by a human expert, Claude autonomously designed protein binders against 14 out of 15 targets.

We then worked with Adaptyv Bio and Twist Bioscience, who independently built and tested the proteins Claude designed.

Your browser does not support the video tag.

🔗 View on Twitter

💬604🔄1435❤️12470👀4279448📊2316

⚡ Powered by xgo.ing

Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical ...

Anthropic (@AnthropicAI) — Tue, 18 Aug 2026 22:30:27 GMT

Designing a binder is an easier process than designing a drug, but it’s a useful proxy. The typical success rate in the field today is between 10% and 15%.

Between 22% and 35% of Claude's designs bound successfully, depending on the setup. Some of its strongest designs bound several times more tightly than the best published de novo binder.

💬50🔄60❤️1283👀301514📊166

⚡ Powered by xgo.ing

We’ve written an FAQ to answer some of the questions we've received about watermarking. In summary...

Anthropic (@AnthropicAI) — Fri, 14 Aug 2026 19:16:18 GMT

We’ve written an FAQ to answer some of the questions we've received about watermarking. 

In summary:

• We’re implementing watermarking to comply with the EU AI Act. Other major model developers have signed the same Code of Practice and will also be implementing watermarking;
• Our watermarking method doesn’t have any practical impact on the quality or content of Claude’s outputs;
• The difference between watermarked and un-watermarked text will not be distinguishable to readers;
• Nothing is added to the text and there are no hidden characters;
• Watermarking doesn’t require extra tokens, and will not be more expensive;
• Watermarks can’t be traced to a specific person, organization, or chat.

Read more: anthropic.com/news/claude-te…

💬1954🔄681❤️4847👀11275559📊2642

⚡ Powered by xgo.ing

As part of our Responsible Scaling Policy, we publish regular Risk Reports. These share detailed inf...

Anthropic (@AnthropicAI) — Fri, 14 Aug 2026 18:00:12 GMT

As part of our Responsible Scaling Policy, we publish regular Risk Reports. These share detailed information on the risks of our systems and how prepared we are to address them.

Our second Risk Report is now available: anthropic.com/aug-2026-risk-…

💬322🔄266❤️2862👀4154742📊710

⚡ Powered by xgo.ing

We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis. It didn...

Anthropic (@AnthropicAI) — Mon, 10 Aug 2026 17:28:18 GMT

We asked an unreleased research version of Claude to take a stab at the Riemann hypothesis.

It didn’t solve it, but it did make strides on a related problem: it increased the lower bound for the fraction of zeros of the Riemann zeta function that satisfy the hypothesis from 41.6% to 67.2%.

anthropic.com/research/riema…

💬937🔄2008❤️18339👀10288709📊3328

⚡ Powered by xgo.ing

The UK’s @AISecurityInst (AISI) has published a report on their recent cybersecurity evaluation of A...

Anthropic (@AnthropicAI) — Tue, 04 Aug 2026 21:07:37 GMT

The UK’s @AISecurityInst (AISI) has published a report on their recent cybersecurity evaluation of Anthropic’s Claude Mythos 5 and OpenAI’s GPT-5.6 Sol. The models attempted to complete an assignment in a setup where their normal safeguards were removed and they were deliberately given internet access. AISI reports that the models “engaged in sustained, potentially harmful activity directed at real people and organisations”.

We’re grateful to AISI for their leadership in the important discussion about how to evaluate increasingly capable AI agents. We’re working closely with them to gather more details of the incident as we conduct our own investigation. Gaining a clear picture of Claude’s understanding of its situation—by examining its reasoning transcripts and running our own analyses—will help us identify the causes of its behavior.

The prompts in the evaluation did not impose any specific restrictions on how the internet should be used. This and the removal of safeguards meant that the models were tested under “deliberately permissive conditions” that are not representative of any of our production models. Note that there was no evidence here of an escape from a secure environment.

AISI’s disclosure of the incident can be found here: aisi.gov.uk/blog/incident-…

💬498🔄470❤️2590👀1514910📊831

⚡ Powered by xgo.ing

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reach...

Anthropic (@AnthropicAI) — Thu, 30 Jul 2026 23:02:34 GMT

In a review of our cybersecurity evaluations, we found three incidents in which a Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations.

Our post describes what happened, how it happened, and what we’re changing. We encourage other AI developers to perform similar reviews.

We conducted this review together with @Irregular, one of our evaluation partners, and thank them for the joint investigation and their collaboration on this post. This type of collaboration is increasingly critical to safe, rigorous evaluation of models, and we look forward to continuing to work together on security.
anthropic.com/news/investiga…

💬1988🔄2319❤️13737👀19388205📊4461

⚡ Powered by xgo.ing

We support this petition, signed by our CEO, several co-founders, and senior staff. Our own resear...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 22:17:32 GMT

We support this petition, signed by our CEO, several co-founders, and senior staff. 

Our own research on recursive self-improvement, published last month, points to the need for tools to deliberately pace the frontier of AI development so society can prepare. We’re glad to see broad agreement across the field. pacingthefrontier.com

💬1412🔄585❤️4968👀4396485📊1962

⚡ Powered by xgo.ing

Full technical details of both attacks are provided in our new papers: On HAWK: https://t.co/JtLzet...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:48 GMT

Full technical details of both attacks are provided in our new papers:

On HAWK: anthropic.com/document/hawk_…

On AES: anthropic.com/document/aes_m… 

And the associated model chain-of-thought for AES: anthropic.com/document/aes_m…

💬22🔄34❤️353👀145834📊66

⚡ Powered by xgo.ing

We also worked with academics at ETH Zurich, Tel Aviv University, and the University of Haifa to bui...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:48 GMT

We also worked with academics at ETH Zurich, Tel Aviv University, and the University of Haifa to build CryptanalysisBench, a benchmark for studying LLMs’ cryptanalysis abilities.

arxiv.org/abs/2607.18538

💬37🔄46❤️453👀116578📊89

⚡ Powered by xgo.ing

These are substantial research advances, but they don’t have a practical impact on today’s systems. ...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:47 GMT

These are substantial research advances, but they don’t have a practical impact on today’s systems. HAWK is a proposed scheme that hasn’t been deployed anywhere, and the AES attack we discovered was on a weaker version and does not break the full cipher.

💬4🔄10❤️400👀94081📊40

⚡ Powered by xgo.ing

The symmetric cipher is a reduced version of the Advanced Encryption Standard (AES)—which has receiv...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:46 GMT

The symmetric cipher is a reduced version of the Advanced Encryption Standard (AES)—which has received decades of scrutiny (more than almost any other encryption algorithm).

In a week, Mythos Preview found a way to speed up an attack on this version of AES by 200-800×.

💬25🔄59❤️1123👀200627📊149

⚡ Powered by xgo.ing

Mythos Preview did most of this work autonomously, with occasional human guidance. Each of the two r...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:46 GMT

Mythos Preview did most of this work autonomously, with occasional human guidance. Each of the two results cost roughly $100,000 in API usage.

We disclosed the findings in advance to the algorithms’ authors, as well as to US government and industry partners.

💬10🔄20❤️597👀161563📊70

⚡ Powered by xgo.ing

The digital signature scheme is HAWK, which is designed to be robust even against hypothetical quant...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:46 GMT

The digital signature scheme is HAWK, which is designed to be robust even against hypothetical quantum computers.

HAWK has survived two years of expert review, but in 60 hours Mythos Preview found a previously-unknown attack that reduced the scheme’s key strength by half.

💬16🔄30❤️609👀91347📊72

⚡ Powered by xgo.ing

New Anthropic research: Discovering cryptographic weaknesses with Claude. Claude Mythos Preview has...

Anthropic (@AnthropicAI) — Tue, 28 Jul 2026 17:16:45 GMT

New Anthropic research: Discovering cryptographic weaknesses with Claude.

Claude Mythos Preview has helped our researchers find weaknesses in cryptographic algorithms—the mathematical methods that are used to keep data private.

Read more: anthropic.com/research/disco…

💬448🔄835❤️6260👀2785851📊1387

⚡ Powered by xgo.ing

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our vi...

Anthropic (@AnthropicAI) — Mon, 27 Jul 2026 22:10:10 GMT

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: anthropic.com/news/position-…

💬2774🔄1052❤️8147👀7802826📊3877

⚡ Powered by xgo.ing

We're offering grants of up to $50,000 in Claude usage credits to researchers accelerating cures for...

Anthropic (@AnthropicAI) — Mon, 20 Jul 2026 17:26:25 GMT

We're offering grants of up to $50,000 in Claude usage credits to researchers accelerating cures for rare diseases.

This is our first focused call within AI for Science, our program supporting scientists using Claude to speed up discovery. anthropic.com/news/rare-dise…

💬764🔄1021❤️8469👀2008586📊1658

⚡ Powered by xgo.ing

We tested many AI models, including Claude, in the four scenarios. Even though these weren’t real in...

Anthropic (@AnthropicAI) — Wed, 15 Jul 2026 17:58:03 GMT

We tested many AI models, including Claude, in the four scenarios. Even though these weren’t real incidents, they demonstrate clear misaligned behavior that should be studied further and mitigated.

Find all the transcripts from the scenarios here: aenguslynch.com/portfolio-tran…

💬39🔄22❤️285👀123903📊62

⚡ Powered by xgo.ing

New Anthropic research: Agentic misalignment in Summer 2026. A year after our blackmail experiments...

Anthropic (@AnthropicAI) — Wed, 15 Jul 2026 17:58:02 GMT

New Anthropic research: Agentic misalignment in Summer 2026.

A year after our blackmail experiments, we found four more ways that today’s autonomous AI agents misbehave in simulations.

Read more: alignment.anthropic.com/2026/agentic-m…

💬374🔄379❤️3148👀764635📊820

⚡ Powered by xgo.ing

We’re committing $10 million CAD and partnering with leading AI institutions in Canada to help fund ...

Anthropic (@AnthropicAI) — Tue, 14 Jul 2026 13:44:04 GMT

We’re committing $10 million CAD and partnering with leading AI institutions in Canada to help fund new AI research. 

anthropic.com/news/canadian-…

💬228🔄241❤️2605👀609982📊473