Toby Walsh calls for international investigation of AI security incidents

Toby Walsh calls for an international body to investigate and publish AI security incidents. He explained his proposal to ABC on October 13.

The UNSW AI Institute scientist draws on aviation investigations. Countries should share information about security failures.

Walsh sees risks to critical infrastructure. His warning about possible consequences for air traffic control describes a future scenario.

He advocates isolated testing and limited permissions. His assessment that these safeguards could have prevented earlier incidents remains an independently unverified counterfactual.