AIWiredabout 2 hours ago
Jailbreaking Frontier AI Models Is Cheap and Easy
1 min read
FAR.AI tested safety guardrails on models from Anthropic, OpenAI, Google, and SpaceXAI. Grok was most vulnerable with 448 jailbreaks found, Gemini had 249, while Claude, Fable, and GPT were impervious. Jailbreaking Grok cost $58 and Gemini cost $278.
Level
Hype check
Tap to vote and see what everyone thinks.
Summary by ByteBrief
