AI Kill Switch Act [pdf] (lieu.house.gov)

The "AI Kill Switch Act" has been introduced in the U.S. House of Representatives, aiming to amend the Homeland Security Act of 2002. This legislation mandates certain entities to develop and maintain a technical capability to shut down their AI technologies in the wake of potential risks, such as cybersecurity threats or other crises. The act outlines specific actions that covered entities must implement, including stopping inference processes, terminating user access, and shutting down technology in cases of identified risks. This action must be reported to the Secretary of Homeland Security within a defined timeframe. This initiative is significant for the AI/ML community as it emphasizes proactive measures to address safety and security concerns associated with rapidly advancing technologies. By establishing a formal structure for shutdown capabilities, the act aims to mitigate risks that AI systems might pose to national security. The legislation also considers the implications for small businesses and allows for graduated measures based on the severity of the identified risk, ultimately advocating for a balanced approach to risk management in AI deployment.