AnalysisPolicyCybersecurity
Jun 12, 8:43 AM
Anthropic disputes claimed jailbreak of Claude Fable 5
Anthropic says the alleged jailbreak by hacker Pliny the Liberator does not bypass core safeguards: the model falls back to a less capable model in sensitive domains like cybersecurity. The company emphasizes extensive red-teaming before launch.
·
Jun 12, 8:43 AM