This allows the more advanced Fable 5 model to handle a broader range of tasks that were previously blocked by overly broad safety filters.
The update is significant because it allows the model to assist with everyday health and educational queries, such as interpreting lab results or explaining medical symptoms, without triggering security blocks.
Anthropic notes that Fable 5 can outperform experts on complex biological tasks, creating a "dual-use" risk where the same capabilities used to develop medical treatments could theoretically be misused to create biological weapons.
By tightening the distinction between benign research and dangerous activity, Anthropic aims to provide healthcare professionals and students with frontier AI access while maintaining a safety margin against catastrophic misuse.
Despite these improvements, Fable 5 still defaults to the older Opus 5 model for high-risk professional tasks involving virology, toxicology, and molecular design.
These areas remain restricted because current AI capabilities could provide a "significant uplift," or unique advantage, to malicious actors seeking to develop novel biological threats.
Anthropic plans to eventually close this gap for legitimate drug development and professional research through "trusted access pathways," which would grant verified experts use of the model's full biological capabilities.