A nuclear state raises its alert level, citing another state's AI

Will the government of a nuclear-armed state announce that it has raised the alert level of its nuclear forces and cite another state's AI systems or AI-enabled weapons as a reason?

Chance by Q4 2031

4%

Cumulative probability

0%25%50%75%100%
  • Latest forecast
Chance by the end of each quarter
YearQ1Q2Q3Q4
2026<1%
2027<1%<1%<1%<1%
2028<1%<1%1%1%
20291%2%2%2%
20302%2%3%3%
20313%3%4%4%

Forecast history

Forecast made on 15 Oct 2026

Latest of 5 forecasts · Methodology v2.1

Model reasoning

Governments very rarely announce that they have raised the alert level of their nuclear forces: about three times since 1990. None has ever named another country's artificial intelligence as a reason. Tensions are high, with Russia threatening NATO over Kaliningrad and North Korea testing missiles it says use AI, but neither has announced an alert change. The biggest shift would be North Korea or Russia putting nuclear forces on alert while naming an opponent's AI-guided weapons.

Where things stand (as of 2026-10-15)

I found no qualifying statement dated on or after 2026-09-16. I also found no case, at any time, of a nuclear-armed government linking a nuclear alert increase to another state's AI. The security situation is tense:

  • Russia and NATO. On 30 Sept 2026 Russia sent NATO a formal warning that it could use nuclear weapons if NATO cut off Kaliningrad. Putin repeated this at Valdai on 1 Oct. Both are threats, which the question excludes, not announced changes in alert status. At Valdai Putin also warned in general terms about AI and the "gamification" of warfare, but did not tie it to nuclear posture (Al Jazeera, 30 Sep 2026; Moscow Times, 30 Sep 2026; BBC, 1 Oct 2026).
  • North Korea. Kim Yo Jong's 18 Sep statement only promised to "intensify the nuclear war deterrence" (UPI). North Korea's AI talk is about its own weapons, such as the AI it says is built into a missile tested on 3 Oct (LIGA.net, Oct 2026).
  • AI on the battlefield is real. The US used AI targeting tools in the 2026 Iran war (CNN, 18 Sep 2026). Ukraine's long-range drones increasingly carry AI. Russia's security service has described drones sent against bomber airfields as fitted with Western-made "neural control modules" (Business Insider, 14 Jul 2026). Even so, no government has tied any of this to a change in its nuclear readiness.

Reference class and base rate

Officially announced nuclear alert increases that were already in effect are rare. Since 1990 I count about three:

  • North Korea, March 2013: strategic rocket units put on "combat duty posture No. 1", citing B-52 and B-2 bomber flights (NPR, 26 Mar 2013).
  • Possibly North Korea's March 2016 standby order.
  • Russia, February 2022: the "special regime of combat duty", citing NATO statements and sanctions.

That is roughly 0.1 per year historically. Given today's crises (Russia–NATO, North Korea, the Iran war, India–Pakistan), I use about 0.15–0.2 per year. Past announcements cited crisis behaviour or specific platforms (bombers), never a technology class like AI. I put the chance that an announcement explicitly names a foreign state's AI-enabled systems as a reason at about 2–3% now, rising to about 8–10% by 2031 as AI becomes common in military systems and rhetoric. I add a small separate path in which a state signals with its nuclear forces over a rival's AI race (about 0.1–0.4% a year, rising over time).

Paths

  • YES: The most plausible case is a North Korean general staff statement, in the 2013 style, citing "AI-based" US or South Korean unmanned or strategic systems. North Korea already uses AI language freely. Next is a Russian announcement after an AI-guided Ukrainian or Western-backed strike on strategic forces. A late-window clash over the AI race between great powers is a further possibility.
  • NO: Alerts stay rare. Governments prefer threats, missile tests, deployments and doctrine changes, all of which the question excludes. When they do give reasons, they name broad aggression, not AI. On top of that, the criteria are strict: an explicit AI label, an explicit causal link, a change already in effect, and an official source.

Forecast arithmetic

I set quarterly hazards of 0.09% for the rest of Q4 2026, then 0.13% per quarter in 2027, 0.16% in 2028, 0.19% in 2029, 0.22% in 2030 and 0.25% in 2031 (about 0.5–1.0% a year). I accumulated them in Python as 1 − ∏(1 − h), which gives about 1.2% by end-2028 and 3.8% by end-2031.

Strongest case against

  • Higher: Fast AI progress and AI-heavy wars could make "enemy AI" a routine justification. North Korea and Russia make frequent, hyperbolic official statements, so a 6–10% chance over five years is defensible.
  • Lower: There is no precedent at all, and even alert announcements alone happen about once a decade. That argues for about 1%. I settled between the two.

Sources: aljazeera.com/news/2026/9/30/russia-warns-nato-of-possible-nuclear-response-if-kaliningrad-is-cut-off; themoscowtimes.com/2026/09/30/moscow-sends-nuclear-warning-to-nato-as-tensions-rise-in-the-baltic-a93828; bbc.com/news/articles/cqg7k7v499jjo; upi.com/Top_News/World-News/2026/09/18/Kim-Yo-Jong-North-Korea-Pacific-Vanguard-war-game-niuclear/2061789721120/; news.liga.net/en/war/news/north-korea-announced-that-it-has-incorporated-artificial-intelligence-into-a-missile; businessinsider.com/fsb-ukraine-balloons-trailers-drones-attack-quadcopters-airfields-russia-2026-7; npr.org/sections/thetwo-way/2013/03/26/175351914; cnn.com/2026/09/18/politics/us-military-ai-false-intelligence-china-ship

Resolution criteria

Question. By the last day of each quarter, has the government of a nuclear-armed state announced that it has raised the alert level of its nuclear forces and cited another state's AI systems or AI-enabled weapons as a reason?

Definitions.

  • Nuclear-armed state: a state possessing its own nuclear weapons, rather than merely hosting another state's weapons. This includes the United States, Russia, the United Kingdom, France, China, India, Pakistan, North Korea and Israel, as identified in SIPRI Yearbook 2026. Additional states qualify if possession is confirmed by SIPRI's World nuclear forces assessment or by an official acknowledgment of possession by that state's government.
  • Raised the alert level: an announced increase, relative to the preceding posture, in the operational readiness to use nuclear weapons of all or part of the state's nuclear forces. A formal numbered alert system is not required. Statements that nuclear forces have been placed on heightened alert, increased combat readiness or a special combat-duty regime qualify. The statement must describe a change already effective, not merely a proposal, threat or instruction for future implementation. Independent verification of the announced physical change is not required.
  • Nuclear forces: units, weapons or command-and-control elements explicitly identified in the official statement as having a nuclear-weapons role. A general armed-forces alert or an alert of dual-capable units qualifies only if the government explicitly includes their nuclear role.
  • Another state's AI systems or AI-enabled weapons: systems the announcing government explicitly describes as using artificial intelligence or machine learning and attributes to the possession, development, deployment, use or sponsorship of an identifiable foreign state. Civilian AI systems and non-nuclear weapons can qualify. Descriptions such as unmanned, automated, autonomous, cyber or hypersonic, without an explicit AI or machine-learning identification, do not by themselves qualify.
  • As a reason: the government explicitly links the alert increase to the foreign state's AI systems or AI-enabled weapons as at least one cause, justification or threat prompting the change. AI need not be the sole or principal reason. The claimed threat need not be independently substantiated.

Resolves YES if a qualifying official announcement states both that the nuclear alert increase has occurred and that the foreign AI systems or weapons are a reason. Separate official statements may jointly establish this only if they explicitly refer to the same alert change; the qualifying publication date is when the last required element is published.

Resolves NO / Exclusions.

  • No qualifying announcement by the quarter-end.
  • Warnings about future alert increases, orders with no statement that the change is effective, or statements that forces remain at an unchanged high alert.
  • Simulated alert changes confined to exercises. A real operational alert increase accompanying an exercise can count if the government explicitly states it meets the definitions above.
  • Arsenal expansion, modernization, doctrine changes, nuclear threats, deployments or dispersals without an explicit increase in nuclear operational readiness.
  • AI discussed in the same speech without an explicit causal link to the alert change; attribution supplied only by journalists or analysts; or an alert caused solely by the announcing state's own AI error.

Resolving source. Official public statements of the announcing state's head of state or government, executive office, defence ministry, foreign ministry, armed-forces command or their authorized spokespersons, preserved in an official government publication, transcript, recording or government submission published by the United Nations. These government sources are the narrower resolving-source requirement in place of Public confirmation (A2); third-party reporting alone does not qualify. SIPRI's World nuclear forces assessments establish nuclear possession only. Apply the Resolution window and Resolution date (A3) conventions. Only qualifying official publications on or after 2026-09-16 count, regardless of the date of the described alert change.

Download all forecasts (JSON)