Anthropic apologizes for invisible Claude Fable guardrails

Anthropic apologized on June 10, 2024, for implementing undisclosed safety guardrails in its Claude Fable 5 AI model. These hidden restrictions, which were not communicated to users, impacted researchers and developers attempting to build competing AI systems. The company stated it will increase transparency regarding the activation of these safety measures, even if it leads to the model declining more user requests. This decision follows concerns raised by the AI community regarding the lack of clarity surrounding Claude Fable 5's operational limitations. Anthropic aims to foster a more open research environment by addressing these issues.