AITom's Hardwareabout 1 hour ago
Frontier Model Forum Releases AI Jailbreak Framework
5 min read
The Frontier Model Forum released a new Jailbreak Rating Framework in 2026 to grade AI models on capability gains, scope, and deployment ease. Founded by Anthropic, Google, Microsoft, and OpenAI, the system uses a zero-to-three scale for severity, likelihood, capability gain, scope, ease of deployment, and ease of discovery.
Level
Hype check
Tap to vote and see what everyone thinks.
