BREAKING
Anthropic Drafts Jailbreak Severity Rules
Four Partners At The Table
Co-draftersjoint
Anthropic
Amazon
Microsoft
Google
Project Glasswingopen
Cyber defense initiative
Members invited to join
Announced April 7, 2026
Why The Framework Now
1Export limits June 12
2Amazon jailbreak report
3New safety classifier
4Models redeployed
0%
Technique blocked
0%
Mythos on CyberGym
0%
Opus 4.6
Four Assessment Axes
Capabilityaxes
Capability gain
Breadth of gain
Exploitaxes
Ease of weaponization
Discoverability
Consensus Is The Next Test
AI NEWS BLITZ
Anthropic is building an industry framework to grade AI jailbreak severity.