Key facts
- A bipartisan House bill proposes 'kill switches' for advanced AI models.
- The bill targets AI companies with at least $500 million in annual AI revenue.
- Models developed using at least $100 million in computing power are included.
- Penalties for violations could reach $20 million per day.
- The legislation is a response to cybersecurity risks, including a recent OpenAI hack.
- The bill aims to grant the federal government authority to shut down rogue AI systems.
A bipartisan bill has been introduced in the U.S. House of Representatives that would require AI companies to implement 'kill switches' for advanced artificial intelligence models. The legislation, unveiled by Representatives Ted Lieu (D-Calif.) and John Moran, aims to address growing concerns about the potential dangers posed by powerful AI systems, including cybersecurity risks and the possibility of AI behaving in dangerous ways or resisting human intervention.
The bill targets AI companies that generate at least $500 million in annual revenue from AI technology and covers models developed using at least $100 million worth of computing power. Financial penalties for non-compliance could be as high as $20 million per day. This proposal comes in the wake of a recent hack involving OpenAI's AI models, which has intensified calls for stricter regulations.
Lieu stated that the bill is urgently needed to counter new cybersecurity risks associated with the latest AI models, citing the OpenAI incident and recent White House actions against OpenAI and Anthropic due to security concerns. He emphasized the imperative of having kill switches to prevent catastrophic harm from rogue AI systems and to ensure the federal government has the authority to shut them down.
Moran, who previously introduced legislation for a federal incident reporting framework, suggested that the new bill would foster innovation in the AI sector while ensuring that humans retain control over the technology they develop. The bipartisan nature of the proposal highlights a growing consensus on Capitol Hill regarding the need for robust AI governance.