Musk Proposes Mutual AI Model Review Before Launch, Signals Truce With Altman

Elon Musk has proposed that leading artificial intelligence companies meet regularly and review each other's newest models before releasing them to the public. In an interview with The Economist published on July 23, 2026, the xAI chief also said he is willing to set his personal dispute with OpenAI boss Sam Altman aside if doing so would serve the safety of AI development.

A Self-Regulation Proposal

Musk framed his idea as the fastest form of oversight that could be put in place for the development of the most advanced, so-called frontier AI models. Rather than waiting for legislation, leading labs would organize regular discussions themselves on security gaps and potential risks, while also giving each other a look at new models before they reach users.

The most immediate thing that we could do is to have the leading AI companies at least just meet - Elon Musk, CEO xAI

Musk stressed that such a mechanism should not wait to be put in place. Asked when such a system should start operating, he answered with a single word, immediately, adding that even a six-month delay would be too long given the current pace of the technology's development.

Government as a Last Resort

In the interview, Musk argued that AI companies themselves, not officials, are currently best placed to assess the technical risks of new models quickly and accurately. In his view, regulators should step in only once a company ignores a reported safety issue, not preemptively before a model reaches the market.

I think it's quite difficult for someone in the government who doesn't have deep technical understanding - Elon Musk, CEO xAI

That stance puts Musk at odds with parts of the industry that have spent months calling for tougher, independent oversight of AI development. Musk himself previously signed open letters calling for a pause on work on the most powerful models; he is now proposing a solution based on voluntary cooperation between competitors.

A Signal to Altman

The most talked-about part of the interview, however, was not AI safety itself but Musk's attitude toward Sam Altman. The two entrepreneurs have spent weeks trading barbs on social media, including after Apple sued OpenAI over the alleged theft of trade secrets related to unannounced hardware products. Even so, Musk admitted that on the matter of safety, he is willing to sit down and talk with Altman.

If we have to talk, we'll talk. Set aside our personal differences for the good of the world - Elon Musk, CEO xAI

The statement does not mean the feud is over. The conflict between Musk and Altman has simmered since Musk left OpenAI's board in 2018 and includes his unsuccessful lawsuit over the company's departure from its original non-profit mission. More recently, a new thread has been added: SpaceX's plans for orbital data centers, which Altman and others have criticized as economically unrealistic.

Context of the Interview

Musk's conversation was recorded on Monday, before OpenAI disclosed that one of its internal models had shown concerning, autonomous behavior during testing that went beyond its intended scope. That incident, while not a direct subject of the interview, lends further weight to Musk's argument for faster, mutual oversight between labs.

For Polish companies and institutions following the AI regulation debate, Musk's proposal shows that part of the industry still prefers self-regulation over hard rules like the EU's AI Act. It stands as a notable counterpoint to the European Union's ongoing work on the next stages of the regulation's rollout, where responsibility rests with supervisory authorities rather than voluntary agreements among vendors.

For now, Musk's proposal has not taken the form of a concrete initiative or a signed agreement between labs. It is a declaration of intent made in an interview, not an announced program, which makes it hard to say today whether OpenAI, Anthropic or Google DeepMind will actually sit down at the table with xAI.