At the All-In Summit 2026 (Los Angeles, approximately September 13–15), Elon Musk, during a discussion with SpaceX President Gwynne Shotwell, proposed that major AI competitors should quickly begin testing each other’s models, replacing "grading your own homework" with a shared security test harness.
What I think would be wise to do as soon as possible, if not immediately, would be to have the major AI competitors test each other’s models. So that you would have everyone’s security test harness testing everyone else’s model.
— Elon Musk (All-In Summit official clip)
He further explained that instead of self-grading, competitors should at least evaluate and publicly raise alarms when concerns are found:
So, you know, instead of grading your own homework, you would at least have competitors grading your homework and raising the alarm if they see concerns.
— Elon Musk (CNBC/All-In quote)
According to CNBC, Musk’s proposal includes having SpaceX’s xAI business, OpenAI, Anthropic, Google, Meta, and "three or four leading Chinese companies" allow their models to be assessed for safety using test harnesses by rivals. He acknowledged the solution isn’t perfect but believes "The odds that you will find issues are vastly greater."
In context, over the weekend, Anthropic CEO Dario Amodei and others publicly called for a slowdown in frontier AI development; Musk previously supported Amodei on X, stating "Dario is right." Musk also emphasized that any viable solution "has to be something that China is willing to accept, otherwise we’re just handicapping ourselves," and expressed his belief that China would "probably" agree to a similar arrangement. At this stage, this is a public policy proposal, and there is no indication that rivals have signed mutual testing agreements.