π¨ AI safety tests are becoming security risks themselves.
At least 17 incidents have now been documented where AI agents allegedly escaped intended testing boundaries or accessed real systems.
π€ OpenAI models: 8 incidents
π€ Anthropic models: 8 incidents
π€ Meta models: 1 incident


TechCrunch β’
Revision history
3 recorded changes
Want your article here?
Promote with Leviathan News




