Developing Nuclear Safeguards for AI
… This demonstrates a model of public-private partnerships that can be replicated in other national security domains. …
Search
Redeploying Claude Fable 5
… These prodigious cybersecurity capabilities make it uniquely attractive to malicious actors who wish to misuse it in cyberattacks. …
Mapping AI-enabled cyber threats: Insights from the LLM ATT&CK Navigator
… It also serves as a guide to how threat actors are likely to misuse increasingly capable models in the near future, giving defenders a chance to get ahead of them. What we learned from this and other analyses directly shapes how we build Claude to prevent such misuse. …
LLMs and biorisk
… This monitoring provides insight into patterns of misuse and helps us determine if and when to take enforcement action. National action We are heartened to see investment in biosecurity as a key component of the new US AI Action Plan . …
Claude Fable 5 and Claude Mythos 5
… Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage. …
Project Glasswing: An initial update
… We’ve also begun our Cyber Verification Program , which allows security professionals using our models for legitimate cybersecurity purposes such as vulnerability research, penetration testing, and red-teaming to do so without certain safeguards designed to prevent cyber misuse. …
Inviting hard questions
… That public benefit mission has led us, for example, to invest in AI safeguards to reduce the risk of misuse, to research the behavior and inner workings of AI models to help us align them to beneficial goals, to give our AI models for free to scientists , and to launch a fellowship program that pa… …
Partnering with Mozilla to improve Firefox’s security
… From model evaluations to a security partnership In late 2025, we noticed that Opus 4.5 was close to solving all tasks in CyberGym , a benchmark that tests whether LLMs can reproduce known security vulnerabilities. …
Building AI for cyber defenders
… We believe we are now at an inflection point for AI’s impact on cybersecurity. For several years, our team has carefully tracked the cybersecurity-relevant capabilities of AI models. …
Reverse engineering Claude's CVE-2026-2796 exploit
… Other pages including about:home have Content-Security-Policy headers that block WebAssembly execution. …