Elon Musk’s bold AI safety idea: 'Let rivals test each other’s models'

The tech landscape is swarming with a plethora of AI fears and warnings, coming from researchers, the international community and tech giants. Everyone is worried about the potentially destructive impacts of AI on humanity. The most terrifying is the extinction of humans at the hands of artificial intelligence.

Against the backdrop of a frightening environment shrouded by clouds of fear and uncertainty, the tech moguls including Nvidia Jensen Huang attended the All-In Summit.

Elon Musk attended the summit virtually and proposed a framework for frontier AI models safety.

The CEO of SpaceX proposed that leading AI companies test one another’s models before release, so competitors can flag safety concerns. Musk said this peer-review system could begin immediately.

“All the AI companies have a test harness, a series of tests that you give to any model to see if it’s going to build bioweapons or nuclear bombs or be deliberately deceptive. Everyone applying everyone else’s test harness is probably the best thing we can do to ensure our safety. We should try to do that as soon as possible,” Musk proposed.

Musk continued that instead of grading your own homework, you would at least have competitors grading your homework and raising the alarm if they see concerns.

According to Musk, the onboarding of China on this framework would be important for its efficacy. “We should probably get an agreement with China. It would be peer review, where the leading AI companies all test each other’s models before release.”

The billionaire also believes that one basically provides API access in advance of the model release. If a tech firm doesn't solve what’s problematic, competitors can go public with the fact that they think the model being released is unsafe. The legal liability would be enormous.

Musk also highlighted that when it comes to safety, Anthropic puts more care into it than OpenAI.

“On balance, I think Anthropic puts more care into their safety than OpenAI. But even Anthropic acknowledges they are worried about their models,” he added.