Policy task forces proposed criminal liability for tech executives deploying uncontained AI systems that cause cyber breaches. Civil fines get treated as operating costs. Legal liability forces leadership to enforce hard containment sandboxes before deploying autonomous agents.