Experts Call for Regulation of AI Compute Hardware to Ensure Safety

A new report highlights the need for policies targeting AI's physical components to prevent misuse and enhance governance.

A new report co-authored by experts from the University of Cambridge and OpenAI emphasizes the need for AI compute hardware regulation to prevent misuse and enhance safety.
The report suggests measures such as a global registry for AI chips, compute caps, and distributing a start switch for AI training to multiple parties.
Government efforts worldwide are increasingly focusing on the regulation of computing hardware to govern AI, with policies like the US Executive Order on AI and the EU AI Act.
The report highlights the risks of unregulated AI, including threats to privacy, negative economic impacts, and the potential for centralization of power.
Proposed solutions include kill switches for AI hardware and prioritizing compute for research beneficial to society.