Confluence of Concerns about the Risk of Nuclear War
In a previous briefing (Risk Frontiers, 2020), we described the fundamental threats to humanity analysed by Ord (2020). He stated that we live at a critical time for humanity’s future. He estimated that in the last century we faced a one-in-a-hundred risk of human extinction, but that we now face a one-in-six risk this century. He concluded that most of this increase in risk came from anthropogenic (i.e., human-induced) sources, and that among them, sources other than nuclear war and climate change were dominant.
In previous work, Snyder-Beattie et al. (2019) estimated an upper bound for the background rate of human extinction due to natural causes. Beckstead et al. (2014) addressed unprecedented technological risks of extreme catastrophes, including synthetic biology, geoengineering (employed to avert climate change), distributed manufacturing (of weapons), and Artificial General Intelligence (AGI); see also Hawking (2010). The conclusions of these studies are summarised and the various potential causes of human extinction are ranked in Table 1, expressed as probabilities of occurrence over a specified future time interval.
The five lowest-ranked sources in Table 1 are all from natural risks, including asteroids and comets, supervolcanic eruptions and stellar explosions. These are estimated to collectively contribute a one-in-a-million chance of extinction per century.
Among the seven highest-ranked sources in Table 1, which are all anthropogenic risks, the most obvious risk to human survival would seem to be that of nuclear war, and we have come near nuclear war, mainly by accident, on several occasions. However, Ord doubts that even nuclear winter would lead to total human extinction or the global unrecoverable collapse of civilisation. Similarly, Ord considers that while climate change has the capacity to be a global calamity of unprecedented scale, it similarly would not necessarily lead to human extinction. He also considers that environmental damage does not show a direct mechanism for existential risk. Nevertheless, he concludes that each of these anthropogenic risks has a higher probability than that of all natural risks put together (one-in-a-million per century).
Newly developing risks that Ord considers include pandemics, “unaligned artificial intelligence” (superintelligent AI systems with goals that are not aligned with human ethics), “dystopian scenarios” (“a world with civilisation intact, but locked into a terrible form, with little or no value”), nanotechnology, and extraterrestrial life. The top ranking of Unaligned AI seems prescient given the rapid development of AI in the past five years.
Table 1. Ranking of Risks of Human Extinction
Risk | 1/Probability | Source |
Unaligned AI | 1 kyr | Ord, 2020 |
Engineered Pandemic | 3 kyr | Ord, 2020 |
Unforeseen Anthropogenic Risks | 3 kyr | Ord, 2020 |
Other Anthropogenic Risks | 5 kyr | Ord, 2020 |
Nuclear War | 100 kyr | Ord, 2020 |
Climate Change | 100 kyr | Ord, 2020 |
Other Environmental Damage | 100 kyr | Ord, 2020 |
Natural Pandemic* | 1 Myr | Ord, 2020 |
Supervolcano Eruption* | 1 Myr | Ord, 2020 |
Flood Basalt* | 32 Myr | Snyder-Beattie et al., 2019 |
Asteroid or Comet Impact* | 100 Myr | Ord, 2020 |
Supernova* | 100 Byr | Ord, 2020 |
*natural hazards |
Although Ord (2020) concluded that most human extinction risk comes from anthropogenic sources other than nuclear war or climate change, it seems possible that the risks due to those two causes may have increased in the past five years. This briefing looks at events that (except for the first one) (occurred within the span of about one month (January 27 to March 3) recently and may have increased the risk due to nuclear war. Interestingly, that risk may have been increased by the confluence of those events with the top-ranked risk in Table 1: Unaligned AI, as suggested by the 17 February event. This briefing aims to be objective and fact-based.
19 November 2024: Formal lowering of Russia’s threshold for using nuclear weapons
President Vladimir Putin formally lowered the threshold for Russia’s use of its nuclear weapons in response to U.S. President Joe Biden’s decision to let Ukraine strike targets inside Russian territory with American-supplied longer-range missiles. The new doctrine allows for a potential nuclear response by Moscow even to a conventional attack on Russia by any nation that is supported by a nuclear power.
27 January 2026: Doomsday Clock moves to 85 seconds to midnight
The Bulletin of the Atomic Scientists advanced the Doomsday Clock (Bulletin of the Atomic Scientists, 2026, to 85 seconds before midnight, citing increasingly aggressive, adversarial and nationalistic postures of Russia, China, and the United States. A year earlier, the clock was advanced to 89 seconds, citing nuclear risk, AI and the climate crisis. The Doomsday Clock is a symbolic, metaphorical timepiece representing how close humanity is to self-inflicted annihilation from nuclear threats, climate change, and disruptive technologies like AI, and is not an objective measure of risk.
5 February 2026: Expiration of the New Start Nuclear Weapons Treaty
Russia and the US are estimated to have 5,580 and 5,225 nuclear weapons respectively (Arms Control Association, January 2025). New Start, which expired on 5 February 2026, restricted the US and Russia to 1,550 deployed nuclear warheads each. The United States has suggested that China join negotiations for a trilateral treaty to replace New Start, China declined that invitation because it says it has a much smaller number of warheads (about 600). China is thought to be rapidly increasing its number of warheads and is expected to have the fissile material necessary for more than 1,000 nuclear warheads by 2030.
14 February 2026: Uncertainty in the reliability of the US nuclear umbrella
In a speech at the Munich Security Conference on 14 February, U.S. Vice President JD Vance told NATO allies that Europe must prioritize its own defense, giving rise to uncertainty in the reliability of the U.S. nuclear umbrella. On 2 March 2026, French President Emmanuel Macron revealed his vision for how France’s nuclear arsenal could bolster the defence of Europe against Russia (Reuters, 2026) which may increase the number of European countries hosting nuclear weapons.
17 February 2026: Artificial intelligences opt for nuclear weapons surprisingly often
As recounted by Stokel-Walker (2026), Payne (2026) found that advanced AI models appear willing to deploy nuclear weapons without the same reservations humans have when put into simulated geopolitical crises. He set three leading large language models: GPT-5.2, Anthropic’s Claude Sonnet 4, and Gemini 3 Flash, against each other in simulated war games. The scenarios involved intense international standoffs, including border disputes, competition for scarce resources and existential threats to regime survival. The AIs were given an escalation ladder, allowing them to choose actions ranging from diplomatic protests and complete surrender to full strategic nuclear war. The AI models played 21 games, taking 329 turns in total, and produced around 780,000 words describing the reasoning behind their decisions. In 95 per cent of the simulated games, at least one tactical nuclear weapon was deployed by the AI models. This raises the question of whether AI can truly comprehend the feelings that real people have about mass death, let alone extinction. Interestingly, this risk stems from a confluence of both the top-ranked anthropogenic risk (Unaligned AI) and the equal-fifth-ranked risk (Nuclear War) in Table 1.
27 February 2026: US agencies stop use of Anthropic technology amid ethics dispute
The US government originally selected Anthropic’s Claude software because of its superior approach to safety. Co-founder Dario Amodei is concerned that in its present condition, Claude may be unreliable, and that in future, AI may escape human control. In recent months this has spurred government objections to the previously agreed-upon prohibition obtained by Anthropic on its use for domestic mass surveillance and for using autonomous weapons systems without intervening human judgment. On 27 February 2026 the President directed all federal agencies to stop using Anthropic technology (The Guardian, 2026). The irony of this decision in light of Payne’s (2026) findings on the trigger-happy use of nuclear weapons by AI in wargames passed unnoticed. The Defence Secretary directed his department to classify Anthropic as a supply-chain risk to natural security, declaring that no contractor could do any business with Anthropic, even if unrelated to the defence department. This may have a significant impact on the functioning of the whole federal government because it would exclude all of the major technology companies in the U.S.
27 February 2026. US armada made a conventional attack on Iran to avert nuclear proliferation
In a continuation of actions by the US over more than two decades beginning in January 2002 following the revelation of secret, undeclared Iranian nuclear facilities at Natanz and Arak by an Iranian exile group, a US armada made a conventional attack on Iran to quash its development of nuclear weapons. The U.S. is reported to have used Anthropic’s Claude to support Operation Epic Fury against Iran, despite the prohibition that had been ordered that day (Axios, 2026).
Implications for Australia and the Asia Pacific
The current trend toward violation of the United Nations charter regarding territorial integrity, together with the demonstrated vulnerability of nations that lack nuclear weapons and uncertainty in the reliability of the U.S. nuclear umbrella, may spur the proliferation of nuclear weapons, which in turn may increase the risk of accidental or intentional nuclear war.
REFERENCES
Arms Control Association (2025). https://www.armscontrol.org/factsheets/nuclear-weapons-who-has-what-glance.
Axios (2026). Axios AM. https://www.axios.com/newsletters/axios-am-f0954cb2-2f31-4426-87fd-050095005344.html?chunk=2&utm_term=twsocialshare#story2/ 1 March 2026.
Beckstead, Nick, Nick Bostrom, Neil Bowerman, Owen Cotton-Barratt, William MacAskill, Seán Ó hÉigeartaigh, and Toby Ord (2014). Unprecedented Technological Risks. https://www.fhi.ox.ac.uk/wp-content/uploads/Unprecedented-Technological-Risks.pdf.
Bulletin of the Atomic Scientists. News conference in Washington DC, 17 January 2026
Hawking S. (2010), Abandon Earth or Face Extinction, Bigthink.com, 6 August 2010
Ord, Toby (2020). The Precipice: Existential Risk and the Future of Humanity. Bloomsbury.
Payne, Kenneth (2026). AI Arms and Influence: Frontier Models Exhibit Sophisticated Reasoning in Simulated Nuclear Crises. arXiv DOI: 10.48550/arXiv.2602.14740
Reuters (2026). Macron to outline nuclear vision amid European unease over US alliance. https://www.reuters.com/world/macron-outline-nuclear-vision-amid-european-unease-over-us-alliance-2026-02-26/
Risk Frontiers (2020). Ranking of Potential Causes of Human Extinction. Briefing Note 415.
Snyder-Beattie, Andrew E., Toby Ord and Michael B. Bonsall (2019). An upper bound for the background rate of human extinction. Nature Reports, https://doi.org/10.1038/s41598-019-47540-7.
Stokel-Walker, Chris (2026). AIs can’t stop recommending nuclear strikes in war game simulations. https://www.newscientist.com/article/2516885-ais-cant-stop-recommending-nuclear-strikes-in-war-game-simulations/ 25 February 2026.
The Guardian (2024). Putin warns US over using long-range missiles by signing new nuclear doctrine. https://www.theguardian.com/us-news/2024/nov/19/first-thing-putin-warns-us-over-using-long-range-missiles-by-signing-new-nuclear-doctrine. 19 November 2024
The Guardian (2026). Trump orders US agencies to stop use of Anthropic technology amid dispute over ethics of AI. https://www.theguardian.com/us-news/2026/feb/27/trump-anthropic-ai-federal-agencies. 27 February 2026
