MIRI Newsletter #126

Announcing: AI StopWatch In our last update, we mentioned we had something new in the works: a dedicated channel for news and analysis about AI. An experiment from the writers and analysts at MIRI, AI StopWatch posts news and commentary seven days a week. You can read our commentary as it’s published via Notes, browse our daily digests, or listen to a podcast version of the day’s coverage. The idea behind StopWatch: people will be more equipped to start good conversations about AI if they know what’s going on. Long-time MIRI readers might be surprised that StopWatch doesn’t cover extinction risk exclusively; our hope is that by sharing MIRI writers’ honest takes on a wider range of issues—jobs, robots, slop, privacy, and more—non–x-risk-aware readers will be more inclined to engage with our perspective on the long-term risks of AI. Learn more about StopWatch and leave us your feedback here. We’ve also curated some favorite StopWatch coverage from May and June: The AI Doc is streaming in the US In March, we asked you to help make the theatrical release of The AI Doc a success. The team behind the film told us it exceeded (in its first 10 days) what the average documentary typically earns in its entire run, selling around 80,000 tickets in that timespan. That said, it certainly hasn’t reached a critical mass, so we’d love your help to get it in front of more people, especially since it is now streaming on Peacock. Please consider: - Posting a review, if you’ve watched the film - Posting a short video encouraging Peacock viewers to check it out - Telling your friends, family, network, etc. to watch it. (If they don’t have Peacock, they can also access it through video-on-demand platforms.) As we mentioned last time, while MIRI was not involved with the making of the film beyond having two staff members interviewed, we see it as an important resource for alerting policymakers and the general public about AI risk. We’ve also written up an FAQ for filmgoers new to the topic, to supplement some of the bits we wish the film had covered in more detail. MIRI media highlights Since our strategy pivot in 2024, MIRI has been working to inform policymakers and the general public about the risks of frontier AI and the need for a global halt. Here’s a roundup of media highlights from the last quarter: - We published an op-ed in The Hill advocating for the US government to lead coordination of a global halt on frontier AI development. - Nate was one of six panelists in the Isaac Asimov debate hosted by Neil deGrasse Tyson, The Rise and Reckoning of AI. A particular highlight was when Neil deGrasse Tyson closed out the debate by emphatically declaring his support for a treaty banning superintelligence! - A clip from Nate’s Harvard Science Book Talk went viral on TikTok. The talk itself also made waves, according to a survey measuring pre and post event views on AI existential risk. - Two long-form interviews from Nate that are worth a watch, with UK-based outlets Novara Media and Americano. - Malo was set to debate Dean Ball at Yale’s Buckley Institute on the question “Should superintelligence be banned?” Due to a travel delay, Dean Ball couldn’t attend so the event was instead turned into a fireside chat. - Two short clips to highlight: Nate with Helen Toner at the 2026 CNBC CEO Council Summit and Malo connecting Mythos to broader concerns about advanced AI in a Fox News segment. New MIRI blogs and shareable graphics We’ve published several new articles on our blog. But first, here are two great graphics we hope you’ll share far and wide: Also see the companion memo, Promising Signals on AI Governance from China by Joe Rogero. MIRI’s Technical Governance Team (TGT) published six new papers, which will be presented at the International Conference on Machine Learning (ICML)’s workshop on Technical AI Governance Research (TAIGR). Several members of TGT will be attending the conference. You can access the full papers here, along with a summary of each. The Technical Governance Team also published two important papers in late 2025 that we’ve now summarized for a more general audience: - An exploration of four possible AI governance scenarios and the risks and benefits of each. In other words: how do we respond to a world where AI capabilities are on track to surpass humans in the near future? Read the governance scenarios summary. - A sample international treaty, with the intent of demonstrating that it’s possible to enact a global halt on frontier AI development. A halt was the only governance scenario in the paper above that did not carry an unacceptably large risk of catastrophic harm. Read the treaty summary. You may have heard about “the Erdős breakthrough”, when an OpenAI model solved a very well-known open math problem that had previously been unsolvable by humans. Here’s Joe Rogero’s memo on what this means for AI capabilities. Other updates & links - If Anyone Builds It, Everyone Dies is currently available in Spanish, Italian, Bulgarian, Dutch, Korean, Portuguese, and Japanese! We’re expecting a number of other translated editions to hit the shelves later this year, including in German and Mandarin. - Rational Animations released a video inspired by Chapter 10 of If Anyone Builds It, Everyone Dies, which focuses on why AI alignment is a particularly difficult challenge. - We’ve received some generous donations that will help us make bigger, bolder plans going forward. Details along with a recap of our last fundraiser here. - There will be a Stop the AI Race march taking place at 12 PM on July 11th at 1455 3rd street in San Francisco. Register here. - On April 30th, Senator Bernie Sanders hosted a livestream forum: “The Existential Threat of AI and the Need for International Cooperation.” During the livestream, Sanders cited a litany of experts’ warnings and called out Congress for not taking action: “One might think that given the very real threat to humanity, countries might come together to regulate this technology through an international treaty, like we did with nuclear weapons at the height of the Cold War. Has that happened? No, it has not. I’m a member of the United States Senate, and I can tell you unequivocally that there has been no serious discussion about this existential threat.” He ended the forum on a more hopeful note: “I think more and more people are becoming sensitive to this issue. And what we’ve got to do is take this issue all over the world and bring countries together. We’ve done it in the past with regard to nuclear weapons. We’ve done it in the past regarding working together on pandemics. We can do it on this.” - Foregone Films released a 16-minute short film Seat at the Table, a fictional film about the real dangers of superintelligent AI. We think it’s worth signal boosting. - Anthropic expressed support for the option to pause AI development in a June blog post: “We believe it would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology.” The statement received strong media coverage, and we think now is an excellent time to contact your political representatives and ask them to voice support for an international halt on frontier AI development. (You may have seen our email about this.) - Lots of other notable things happened in the AI world this quarter. See updates on Mythos, Fable, the new METR report, and the pope’s encyclical on StopWatch. Onward, Alana Horowitz Friedman and Rob Bensinger