Anthropic's new model Fable has been released with guardrails to limit its use for cybersecurity and biology-related tasks, but some cybersecurity researchers are unhappy with the restrictions, finding them too broad and hindering their work. The model's safety measures flag and reject certain requests, and in some cases, silently degrade the model's performance without notification. This has led to frustration among researchers who feel that the restrictions are overly cautious and stifle legitimate research. The restrictions are intended to prevent the development of malware or biological weapons.