Cybersecurity researchers arent happy about the guardrails on Anthropics Fable | TechCrunch
Summary
Anthropic has released Fable, a limited version of its cybersecurity model Mythos, and built in guardrails to block cyber and biology-related prompts. Security researchers are criticizing the restrictions because they also block legitimate tasks like secure code review and basic software engineering work. The model falls back to Claude Opus 4.8 when it hits a guardrail, and the filtering appears to rely heavily on keywords. Anthropic also runs a Cyber Verification Program that gives approved professionals fewer restrictions for cybersecurity use. The story highlights how AI vendors are balancing safety controls with practical enterprise use cases in cybersecurity.
Classifications
industries
No industries detected
applications
Accounting and Taxes
AskAI Classifications
Labels
AI Software
SaaS
Conversational AI
