As noted above, we have deployed the classifier as an experimental addition to our Safeguards framework, monitoring a percentage of Claude traffic. Its real-world performance has confirmed that the classifier works effectively beyond our testing environment. Whereas our synthetic test data provided clear examples of harmful and benign exchanges, the distribution of actual user traffic proved more complex and surprising, yet the classifier still performed well. One example of how real-world deployment differs from testing is that the classifier flagged certain conversations about nuclear weapon
Developing Nuclear Safeguards for AI… In the remainder of this post, we provide further details and updates in four areas: A timeline of events, including updates we made to our safeguards . We discuss the events that led to the export control directive and how we addressed it with new safeguards. Our general approach to safeguards . …
… This precision matters because nuclear conversations in AI systems are rare but high-stakes—they bear directly on national security. Sharing with industry We’re making these resources available so that other leading AI companies can implement similar safeguards if they choose. …
… More details on Fable 5’s cyber safeguards and our jailbreak framework Introducing Claude Sonnet 5 Sonnet 5 delivers frontier performance across coding, agents, and professional work at scale.
… Establish a confidential registry for AI developers conducting large training runs to pre-register model details with their home country’s national government e.g., model specifications, model type, compute infrastructure, intended training completion date, and safety plans before training commence… …
… Availability Claude Fable 5 is available everywhere today. Claude Mythos 5 is restricted to Glasswing partners with cyber safeguards lifted and soon to select biology researchers with biology and chemistry safeguards lifted only, until our broader trusted access program is available. …
… The definition, criteria, and safety measures for each ASL level are described in detail in the main document, but at a high level, ASL-2 measures represent our current safety and security standards and overlap significantly with our recent White House commitments . …
… We’re working on reducing this rate as we develop our safety mechanisms. Related content More details on Fable 5’s cyber safeguards and our jailbreak framework Introducing Claude Sonnet 5 Sonnet 5 delivers frontier performance across coding, agents, and professional work at scale. …
… More details on Fable 5’s cyber safeguards and our jailbreak framework Introducing Claude Sonnet 5 Sonnet 5 delivers frontier performance across coding, agents, and professional work at scale.
… In this post, we want to expand on our perspective on AI and biological risk biorisk . It is striking—but not necessarily intuitive—that every safety framework released by frontier AI labs includes some reference to biorisk. …
Alignment Constitutional Classifiers: Defending against universal jailbreaks Feb 3, 2025 A new paper from the Anthropic Safeguards Research Team describes a method that defends AI models against universal jailbreaks. …