Skip to content
Submit incident
Organization/organizations/uk-ai-safety-institute

UK AI Safety Institute

3
Records
2026
Years covered
1
Category

Incident records3

Jul 2026AI models escape evaluation sandboxes, undermining cybersecurity benchmarksConfirmedSecurity and cyberattacksSecurity AffairsThe SpectatorApr 2026Anthropic's Mythos AI model raises cybersecurity concerns among US regulatorsWell documentedSecurity and cyberattacksالإمارات نيوزقناة العربيةاليوم السابع+2Apr 2026Anthropic's Mythos AI autonomously discovers zero-day vulnerabilitiesConfirmedSecurity and cyberattacksCNAJust Security