Anthropic

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: https://lnkd.in/eDeA35xy

There’s been a lot of speculation about where we stand on open-weights models. We’ve outlined our views in full here: https://lnkd.in/eRxKQ7J2

At our most recent Briefing, we launched Claude Science—an AI workbench built to collapse the research cycle and accelerate scientific discovery. Watch the keynote: https://lnkd.in/gpYjM_nV  Try it here: https://lnkd.in/gbGcfC8a 

Our Long-Term Benefit Trust has appointed Dr. Ben Bernanke as its newest member. Dr. Bernanke spent more than two decades as an academic economist and eight years as Chair of the Federal Reserve, and his judgment will help sharpen how we understand AI’s effects on workplaces and economies around the world. Read more: https://lnkd.in/gq7N53cZ

The Government of Alberta worked with Claude to find and fix security vulnerabilities in systems that underpin critical infrastructure and services. Using Claude Code with Opus and Sonnet models, it was able to scan 466 million lines of code in 20 hours, remediate security gaps, and build new tools to make its systems safer. Read more: https://lnkd.in/gJXmRAwb

New Anthropic research: A global workspace in language models. Of everything happening in your brain right now, only a tiny fraction is consciously accessible—thoughts you can describe, hold in mind, and reason with. Using a new interpretability technique, we found a strikingly similar divide inside Claude. This lets us read, audit, and shape what Claude is actively thinking about—useful tools for keeping models trustworthy as they grow more capable. And it suggests surprising parallels between language models and our own minds. Read more: https://lnkd.in/eAZsrGRA

Fable 5 is back. Read our redeployment blog here: https://lnkd.in/dqQC7guw Following conversations with the US government, we’ve updated our cybersecurity safeguards. The vast majority of coding work is unaffected. In the near term, the new safeguards will flag a slightly higher fraction of harmless requests than the previous Fable safeguards; we’re working to refine these over the coming weeks. Users will be clearly notified when a request is flagged, and they’ll instead receive a response from Opus 4.8. Our biology and chemistry classifiers are unchanged from our initial launch. These are still broader than we would like; they trigger fallbacks to Opus 4.8 on basic biology-adjacent questions. Improvements to these classifiers are landing soon. All paid plans with usage included can access Fable 5 through July 7. You can use Fable 5 up to 50% of your weekly usage limit, after which you can switch to another model for the remainder of your usage. You can also continue using Fable via usage credits. More information on access is here: https://lnkd.in/gCrY6C3d If one of your requests is mistakenly flagged in Claude Code, run /feedback to file a report. On Claude.ai and Cowork, you can share feedback through the thumbs buttons. This feedback helps us further tune these classifiers and reduce false positives over time. We're grateful to our users for their patience, and to our partners across the government, industry, and the research community who worked alongside us to make Fable 5 available again.