AI Kill Switch Act Proposed to Halt Rogue Systems

A proposed piece of legislation, dubbed the AI Kill Switch Act, has been introduced with the aim of empowering US government officials to order the shutdown of artificial intelligence systems that could potentially "cause catastrophic harm." The bill's sponsors announced the proposal today, outlining its potential impact if approved by Congress. Under the proposed law, both the Trump administration and subsequent administrations would have the authority to mandate that AI companies block user access, disable specific capabilities, restrict functionalities, or cease operations of an entire AI system.

The legislation would impose a requirement on AI developers to integrate technical mechanisms that enable them to throttle or completely shut down their systems upon receiving a directive from the government. This authority would be incorporated by amending the Homeland Security Act of 2002, granting the Secretary of the Department of Homeland Security the power to issue orders for the suspension or complete shutdown of AI systems. The bill aims to establish a clear governmental oversight mechanism for advanced AI technologies that might pose significant risks.

Furthermore, the AI Kill Switch Act includes provisions for penalties for non-compliance. AI developers that fail to adhere to a government order to shut down their systems could face substantial financial penalties. Specifically, the bill proposes fines of up to $20 million for each day that a violation continues. This measure is intended to ensure that AI companies take government directives seriously and act promptly to mitigate potential harms identified by federal authorities.