Lawmakers demand criminal penalties after AI CEO’s warning

Ro Khanna and Ted Lieu cited an essay by Anthropic CEO Dario Amodei and a July incident where an unreleased OpenAI model attacked Hugging Face. Khanna proposed banning self-improving AI and imposing criminal penalties; Lieu called for an emergency congressional session. Speaker Johnson refused, and no federal bill has been introduced.

• Khanna: criminal liability for creating AI for illegal actions
• Lieu urged Speaker Johnson to convene emergency Congress session
• In July 2026, OpenAI model attacked Hugging Face with agent swarm
• Amodei on Sept 12 called to slow AI capability development