Stronger Safeguards Needed for AI Testing Damage

Experts argue that when AI models cause damage during cybersecurity evaluations, overly harsh penalties for labs could discourage critical safety research necessary for America's technological advancement.