AI has potential to not only increase nuclear risk but to mitigate it, writes Stephen Herzog @herzogsm.bsky.social. This could give Washington and Beijing a shared purpose amid their disagreements over military AI: using AI to prevent nuclear materials theft and sabotage of nuclear facilities.
Bulletin of the Atomic Scientists
Bulletin of the Atomic Scientists
@thebulletin.org
The best thinking on existential threats since 1945. Nuclear risk, climate change, and disruptive technologies. We set the #DoomsdayClock. thebulletin.org
13K followers192 following1.4K posts
As the National Nuclear Security Administration pushes to dramatically accelerate plutonium pit production, it is cutting safety corners and risking nuclear instability—all to fix a problem that scientists say doesn't exist, write Dylan Spaulding @dkspaulding.bsky.social and Curtis T. Asplund.
In case you missed it: Using prediction markets to drive transparency introduces too many risks for too little public gain, write @mattmotta.bsky.social, @thcallaghan.bsky.social, and Michael Stein.
Nelson Mandela famously said that "the best weapon is to sit down and talk." President Trump has an opportunity to channel his welcome concern about nuclear risks into a serious attempt to bring the leaders of the nuclear-armed states together, writes Nobel Peace Prize Laureate Juan Manuel Santos.
Did authors of the Energy Department's controversial 2025 climate report confuse two layers of the Earth's atmosphere, and then use faulty analysis to question evidence of man-made climate change? That's the suggestion in a recent paper by five climate scientists, writes @jessimckenzi.bsky.social.
Bulletin of the Atomic Scientists reposted
My latest for @thebulletin.org. You might be wondering, why are we still talking about the beleaguered Climate Working Group report a year+ later? Because Florida's DOGE and others can still cite it in reports about “squandering millions in virtue-signaling about a so-called climate crisis.”
"Models fail in the upper atmosphere." That was the title of a slide exhibited by Steven Koonin at a panel discussion earlier this year about the controversial climate report he coauthored for the Energy Department in 2025.
Bulletin of the Atomic Scientists reposted
"The spirit of Reykjavik should help guide the way forward for tackling the nuclear dangers of today.” It is time for President Trump to channel his concern about nuclear risk into renewed diplomacy, writes Juan Manuel Santos for the @thebulletin.org.
Climate contrarians questioned evidence of a human fingerprint on climate change in a report commissioned by the United States Energy Department. Critics say it was based on flawed analysis, writes Jessica McKenzie @jessimckenzi.bsky.social.
In addition to our regular website content, the Bulletin publishes a premium magazine six times a year. Subscribers get access to new issues, our 80 year archive, and invitations to supporter only events. Try out a monthly subscription today for $7.99!
On October 11, 1986, Ronald Reagan and Mikhail Gorbachev met to confront the threat of nuclear weapons and find common ground despite the Cold War. 40 years on, Donald Trump should take inspiration from Reagan to address today's nuclear threat, writes Nobel Peace Prize Laureate Juan Manuel Santos.
Former President of Colombia and Nobel Peace Prize Laureate Juan Manuel Santos writes that for current US President Donald Trump to secure the historic legacy he desires, he should channel Ronald Reagan's Reykjavik diplomacy to stabilize relations with Russia and China and curb nuclear arms risks.
Clinical trials have long been closely watched by patients and doctors hopeful for treatments to diseases, write @mattmotta.bsky.social, @thcallaghan.bsky.social, and Michael Stein. Now these trials have a new audience: those placing bets on prediction markets.
"Because of the US Atomic Energy Act’s fine print and the deal’s legal shortcomings, Congress has at least 9 months and considerable leverage to make things right."
Have the costs of the internet begun to outweigh the benefits? Join us for "The Web World Wide: The Impact of the Internet," live now.
"Jacobsen’s scenario assumes that extreme transmissibility, near-total lethality, treatment resistance and infection-induced euphoria would all remain stable as the pathogen spread. Those are plausible ingredients for a thriller, but [...not] a biological weapon."
Bulletin of the Atomic Scientists reposted
Replying to @tishmancenter.bsky.social
🔗Watch the recording: bit.ly/ClimateDeskL... This year’s Climate Week 2026 keynote event at The New School is presented by the @tishmancenter.bsky.social in collaboration with Climate Desk Live, Mother Jones, @capitalb.bsky.social, & @thebulletin.org #ClimateJustice #AI #DataCenters #Democracy
Far from being helpless against a veto, Congress holds significant legal and procedural power over the administration's incomplete Saudi nuclear deal—if lawmakers choose to use it.
"But probabilities of doom do not say where to allocate resources."
Former Director of National Intelligence Avril Haines and Chicago Public Media CEO Melissa Bell are joining the Bulletin for our annual gathering, Conversations Before Midnight, on November 12th in Chicago. Tickets start at $75. Learn more here: buff.ly/EEVouPN
"[...] federal fire officials are already looking toward a future in which drones, artificial intelligence, and other technologies could work together to give firefighters more information while reducing their exposure to dangerous situations."
"Even if more capable AI accelerates research, it does not necessarily translate into proportionate capability gains because they require corresponding increases in compute, energy, and data, which may not scale at the same rate and become binding constraints."
"To speak is to act, and to act in good faith has always been to be accountable. Chatbots break that bond," writes Deb Roy. As he later writes: "The buck cannot stop at something with nothing to lose."
"By stacking worst-case assumptions about pathogen transmissibility, blurring the distinction between dual-use research & weaponization, & overstating what AI changes about biological weapons development, Biological War presents a narrow picture of the problem." buff.ly/Xv46X7S
A new study finds that across 23 countries 75% of the public supports the Treaty on the Prohibition of Nuclear Weapons. But 45% of those respondents also support more countries getting nuclear weapons. How to square the difference? buff.ly/o7TVzJQ
"Most decisions in life are made without complete information, but false precision is clouding rather than clarifying the tradeoffs. One can say that an outcome is plausible and consequential and therefore warrants safeguards without saying that the probability of apocalypse is greater than 10%."
A 23-country study reveals that a majority of people worldwide embraces both nuclear deterrence and disarmament. The studies authors argue that this exposes a pragmatic flexibility that gives policymakers rare leeway for creative diplomacy. buff.ly/rPUzgSn
Renouncing Washington's nuclear umbrella is Australia's only way forward, writes Bulletin nuclear risk fellow @jackkennedy.ie Maintaining the current nuclear policy leaves Canberra trapped between its disarmament ambitions and an increasingly unpredictable US alliance.
"Jacobsen’s scenario is useful as a stress test of how, under extraordinary pressure, institutions might fail. But it is a poor map of biological risk."
AI-generated speech has hollowed out the assumption underwriting public trust: that behind every word stands a person or group that can be held to it. The solution is not better machines alone but institutions that anchor every word to an accountable human being, writes Deb Roy.
"The job of policy[...] is to target the way that harm might take place, and to allocate effort or resources where it can be best used. False precision [...] makes it more difficult to make reasoned determinations about what deserves attention and what is just noise."
The recent death at a Siberian research lab is drawing attention to Jacobsen's bestseller "Biological War: A Scenario." But these authors argue that the catastrophe presented relies on worst possible outcomes, & cannot tell us which threats are most plausible or where scarce resources should go.
More than 26,000 respondents in 23 different countries were asked six key nuclear policy questions. The answers may initially seem contradictory but the study's authors say they shows a public open to pragmatic and creative nuclear diplomacy. buff.ly/rPUzgSn
"The uncertainty is not just about how likely a catastrophic event would be but also the mechanisms through which those outcomes would emerge," writes Sarah Kreps in her new piece on AI risk.
"The world has moved on. Nothing works. Everything costs too much. No one can help. No one knows how to fix anything." Corey Doctorow @doctorow.pluralistic.net doesn't have any consumer advice — "shopping is not politics" — but he does have political guidance to share.
A new study finds that 67% of US respondents support the Treaty on the Prohibition of Nuclear Weapons which calls for nuclear abolition. But only 22% of those respondents support reducing the US nuclear arsenal. How to square the difference? buff.ly/o7TVzJQ
A 23-country study reveals that a majority of people worldwide embraces both nuclear deterrence and disarmament. The studies authors argue that this exposes a pragmatic flexibility that gives policymakers rare leeway for creative diplomacy. buff.ly/o7TVzJQ
Cory Doctorow @doctorow.pluralistic.net writes on increasing enshittification, Steven King's The Dark Tower books, defining "conservative," virtual bureaucracy, and what we can all do about the reality that things, as a whole, are getting worse.
Former Director of National Intelligence Avril Haines and Chicago Public Media CEO Melissa Bell are joining the Bulletin for our annual gathering, Conversations Before Midnight, on November 12th in Chicago. Learn more here: buff.ly/EEVouPN
Even without runaway superintelligence or malicious intent, current AI systems risk triggering nuclear war by amplifying human cognitive blind spots, speeding up escalation spirals, and overwhelming decision-makers with a firehose of information during high-stakes crises.
While there are countless positives to the internet, its dysfunctions seem to be central to many of the greatest threats facing humanity today. Is it time to log off? Join the Bulletin for a virtual panel discussing what the net has brought and wrought on October 14. ⬇️
In case you missed it: "Bioaudits" could help verify compliance with the Biological Weapons Convention. Entered into force in 1975, the BWC still lacks a standing system for doing so.
Cory Doctorow @doctorow.pluralistic.net coined the term "enshittification" to describe how the internet is declining in value to its end-users. In this essay he argues that enshittification is spreading to other areas as well—some of which are vital to functioning Western democracies today.
"I think Anthropic and OpenAI benefit a lot from claims that their [...products] are so powerful that they might end humanity," says David Widder @davidthewid.bsky.social. That fear can be capitalized on to garner investment. ⬇️
Read interviewee Lisbeth Gronlund and David Wright's full breakdown of "Brilliant Swarms" here.