tech
Cybersecurity researchers aren't happy about the guardrails on Anthropic's Fable
Cybersecurity researchers are complaining that Anthropic's new model Fable has guardrails that are too strict for any cybersecurity work.
TL;DR
- Anthropic released Fable, a public version of its cybersecurity model Mythos.
- Cybersecurity researchers are complaining about Fable's strict guardrails.
- Fable rejects requests tangentially related to cybersecurity, even simple tasks like reading a blog post.
- The guardrails are in place to prevent the model's misuse for developing malware or biological weapons.
- Researchers find the restrictions haphazard and keyword-based, hindering legitimate cybersecurity work.
- Anthropic has a Cyber Verification Program for cybersecurity professionals, which offers fewer limitations.
- OpenAI also has a similar program called Trusted Access for Cyber.