Ro Khanna Introduces Human Control Over AI Act

Representative Ro Khanna introduced the Human Control Over AI Act to regulate artificial intelligence and mitigate what he describes as a "civilizational extinction risk." The proposed legislation seeks to ban recursive self-improving AI and models capable of autonomously modifying their own core objectives or shutdown controls until the government establishes federal safeguards.

The bill proposes the creation of a new federal agency to oversee frontier AI models from companies including OpenAI, Anthropic, Google DeepMind, and xAI. This agency would implement a licensing system, mandate audits by independent embedded auditors, and enforce security standards such as air-gapped testing environments and kill switches. To ensure accountability, the act introduces criminal penalties for employees who disable safety systems and for "crimes against humanity" resulting from AI deployment. AI companies would also be required to maintain extensive liability insurance.

Khanna stated the bill is modeled on input from AI safety nonprofits rather than corporate executives. While the proposal aligns with other legislative efforts like the bipartisan FRONTIER Act, no House votes on these measures are expected until after the midterm elections.