--- title: "Sam Altman Says the Way to Cooperate With Elon Musk Is to Punch Him in the Nose" description: "OpenAI’s chief calls Musk a bully who respects only those who hit back, then promises he will “of course” work with him on AI safety" author: "Nate Ledger" published: 2026-10-05T18:09:53Z modified: 2026-10-05T21:19:33Z url: https://rews.cc/a/sam-altman-says-the-way-to-cooperate-with-elon-musk-is-to-pu-4537a6 language: en tags: ["ai", "openai", "musk", "altman", "safety", "tech"] publisher: "Rews (https://rews.cc)" --- # Sam Altman Says the Way to Cooperate With Elon Musk Is to Punch Him in the Nose *OpenAI’s chief calls Musk a bully who respects only those who hit back, then promises he will “of course” work with him on AI safety* By Nate Ledger · October 5, 2026 · https://rews.cc/a/sam-altman-says-the-way-to-cooperate-with-elon-musk-is-to-pu-4537a6 ## In brief - Altman topped Vanity Fair’s revived New Establishment list and was interviewed by editor Mark Guiducci - He called Musk a bully who only respects people who hit back, but said he will work with him - In July Musk called him “Scam Altman”; Altman mocked Musk’s pitch to investors on space data centers - A jury found Musk’s suit against OpenAI time-barred in May; his lawyer said they would appeal - OpenAI, Anthropic, SpaceXAI, Google and Nvidia signed a voluntary, White House-driven AI safety accord Vanity Fair has revived its New Establishment list for the first time in nearly a decade, and Sam Altman, the co-founder and chief executive of OpenAI, came out on top. Mark Guiducci, the magazine’s editor, interviewed him for the occasion and asked whether the AI industry could “grow up” and work together. Altman’s answer, [as reported](https://www.yahoo.com/news/politics/articles/sam-altman-says-elon-musk-161441808.html) when the interview was published on Monday, started with Elon Musk: “Elon’s a bully.” > I think once in a while you’ve got to punch a bully back in the nose. And I also think Elon, like many bullies, only respects people who do. Then he got to the part that is actually about cooperation: “If you want me to be able to work with him, maybe you should be happy I hit back once in a while, but I will work with Elon, of course.” I like this a lot as a theory of negotiation. The usual view is that insulting your counterparty makes a deal harder. Altman’s view is that insulting this particular counterparty is a *precondition* for any deal, because Musk only deals with people who have shown they will hit back. On this reading, every rude post is a deposit into a credibility account, and the balance gets spent later when everyone sits down to agree on AI safety. Game theorists have a name for the strategy (tit-for-tat). Most people just call it a feud. Guiducci pointed out that Altman’s occasional public “negging” of Musk doesn’t sit easily with the camaraderie the industry is now trying to project. Altman did not accept that the problem was symmetrical. “Who do you think has been nicer to each other in that relationship?” he asked. “I won’t call myself a saint. But I’m not sure if I’m the one that you’re supposed to come after between Elon and me.” Near the end of the interview he came back to it, [according to TheWrap](https://www.thewrap.com/industry-news/tech/sam-altman-elon-musk-bully-ai-working-relationship/): Musk is “not nice,” but Altman is willing to work with him and wants to. ## The record of niceness For a sense of what the two men’s niceness looks like in practice, Business Insider went back to July. Another account posted that Musk had “warned the world about Scam Altman,” which has become Musk’s nickname for his former ally. Musk quoted it and added: “He takes scamming to a whole new level.” Altman responded by going after Musk’s enthusiasm for orbital data centers, which Altman has repeatedly said he doubts: “Homeboy you’re the one selling public market investors on short-term space datacenters.” I have to admire the comeback. Musk’s insult is the old-fashioned one: you are a scammer. Altman’s is the securities-lawyer one: you are telling *public market investors* things about your business that I don’t believe. In business circles that is the sharper accusation, because public investors are the people you are legally obliged not to mislead. “Homeboy” is doing a lot of work to keep it sounding casual. The falling-out has also gone to court, more than once. The biggest case went the way of Altman and OpenAI in May. After a three-week trial in Oakland, California, an advisory jury [unanimously found](https://www.nbcnews.com/tech/tech-news/openai-elon-musk-case-verdict-rcna345655) that the three-year statute of limitations barred Musk’s claims against Altman, co-founder Greg Brockman and OpenAI, and also cleared Microsoft of aiding and abetting. US District Judge Yvonne Gonzalez Rogers adopted the jury’s findings and dismissed the case. Deadline described it as a [$150 billion lawsuit](https://deadline.com/2026/05/elon-musk-verdict-openai-sam-altman-1236914787/) and noted that the jury never got to the merits. Musk’s lawyer, Steven Molo, said in court that they would appeal. The judge observed that Musk may face an uphill battle, since whether the deadline had passed is a factual question for the jury. Some arithmetic: Musk put [$38m into OpenAI](https://www.aljazeera.com/news/2026/5/18/elon-musk-loses-lawsuit-against-openai) in its early years, according to Al Jazeera. $150 billion is about 3,950 times that. Lots of people regret selling early, but few of them sue for a return like that. ## The group chat that isn’t Normally none of this would matter beyond entertainment value. It matters a bit more now because the big AI companies are supposed to be cooperating on safety. Leaders of OpenAI, Anthropic, SpaceXAI, Google and Nvidia signed a White House-driven agreement under which they [take on self-imposed safety measures](https://rews.cc/a/dinner-with-the-doomers-e6af15). Note the word “self-imposed.” A voluntary accord has no regulator to enforce it, so in practice it rests on the signatories watching each other and caring what the others think. One of the signatories has a name that tells you whose company it is, and Altman’s stated view is that its boss respects only people who punch him. So how do the labs actually talk? Altman explained on Politico’s Decoded podcast: “There’s no like big WhatsApp that everybody’s in, but there’s like lots of, you know, people are running into each other at the same conference. There’s calls, occasional meetings, stuff like that.” Hallway chats at conferences, phone calls and the odd meeting: that is the coordination machinery for a technology its own makers say needs careful steering. It is also, more or less, how every industry coordinates, which is either reassuring or not, depending on how special you think this one is. Altman’s idea of where the danger comes from is also worth a look. [The San Francisco Chronicle](https://www.sfchronicle.com/entertainment/movies-tv/article/sam-altman-andrew-garfield-vanity-fair-22463245.php) and others reported that in the same interview he warned of “a coming tidal wave of cybersecurity problems” from rival open-source models, and suggested OpenAI could help its customers defend against it. So the risk comes from other people’s models and the protection comes from an OpenAI product. That fits his [recent habit](https://rews.cc/a/sam-altman-thinks-the-other-ai-lab-is-the-one-with-the-god-p-460250) of finding the safety problem at someone else’s lab. The interview came out the same day an [OpenAI safety leader resigned](https://rews.cc/a/openai-safety-leader-quits-and-warns-the-company-is-not-bein-9598f0) with a warning that speed is outrunning safety at the company. Anyway, the industry now has a voluntary safety pact, no group chat, and a working theory from its best-known chief executive that cooperation with one of the other signatories comes from an occasional punch in the nose. Nobody has written that clause into the accord.