Claude Fable 5 Redeploys Globally with Enhanced Safeguards

Anthropic has announced the global re-release of its Claude Fable 5 model, following a period of productive discussions with the US government. The model will now include an updated set of classifiers specifically designed to identify and block a broader range of cybersecurity-related tasks. Initially, some routine functions like coding and debugging will temporarily revert to the Opus 4.8 model as the new classifiers are refined to minimize false positives and better distinguish legitimate requests from misuse. In a broader industry effort, Anthropic has also initiated the drafting of a consensus framework alongside major partners such as Amazon, Microsoft, and Google. This framework aims to standardize the assessment of AI jailbreak severity and establish best practices for how AI developers should respond to such incidents. Other industry partners and model providers are invited to join this collaborative initiative. Furthermore, Anthropic is scaling up its collaboration with the US government on model testing and safeguards. This partnership will involve providing pre-release access to models for evaluation, sharing information on jailbreaks and misuse patterns, and dedicating resources to joint research efforts. The company expressed gratitude to users and partners for their patience and collaboration in making Fable 5 available again.