Rep. Khanna to introduce AI safety bill banning 'recursive' self-improving AI

Democratic Rep. Ro Khanna will introduce the Human Control Over AI Act, which sets strict liability standards and bans recursively self-improving AI until federal safeguards exist. It would create a new agency overseeing OpenAI, Anthropic, Google DeepMind and xAI, with licensing, audits, kill switches and criminal penalties for disabling safeguards.

• Bans models that recursively self-improve or alter their own shutdown controls
• Creates a federal agency with licensing, audits and safety standards for frontier labs
• Requires independent auditors embedded at every frontier lab reporting to the agency
• Criminal penalties for disabling safeguards and mandatory liability insurance for releases