Call for ‘red lines’ would move global AI safety, security institutes into enforcement role

More than 270 prominent individuals and organizations from industry, academia and civil society are calling on governments to agree on drawing a set of “red lines” around the development and proliferation of artificial intelligence, which they suggest could be enforced in part by an enduring international network of AI safety and security institutes.

“An international agreement on clear and verifiable red lines is necessary for preventing universally unacceptable risks,” reads their call published Sept. 22 during the 80^th session...