Developing Nuclear Safeguards for AI
… We have already deployed this classifier on Claude traffic as part of our broader system for identifying misuse of our models. …
Search
Building safeguards for Claude
… This helps us spot behaviors that might appear violative only in aggregate, such as automated influence operations and other large-scale misuses. …
Inviting hard questions
… But many also hold serious concerns. …
Clio: Privacy-preserving insights into real-world AI use
… Misuse of Clio: A system like Clio could be misused to engage in inappropriate monitoring. …
Societal Impacts Research
… How is AI used and misused in the wild? …
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
… Across all models, our Safeguards team monitors activity on our platforms for signs of misuse, including dangerous applications of biology. This monitoring provides insight into patterns of misuse and helps us determine if and when to take enforcement action. …
Understanding and addressing AI harms
… This analysis helps us develop approaches that maintain the utility of these capabilities while incorporating appropriate monitoring and enforcement to prevent misuse. …
Redeploying Claude Fable 5
… These prodigious cybersecurity capabilities make it uniquely attractive to malicious actors who wish to misuse it in cyberattacks. …
Claude Fable 5 and Claude Mythos 5
… We therefore need strong safeguards to prevent misuse, and their coverage needs to be broad. …