Apart from guardrails inside its models, Anthropic requires cybersecurity professionals to apply to the Cyber Verification Program. If they get approved, the applicants have fewer limitations on using Claude for cybersecurity work. OpenAI has a similar program called Trusted Access for Cyber.


