Photo by imsogabriel stock on Unsplash
BERLIN, GERMANY [TAC] — The collapse of traditional arms control treaties and the integration of artificial intelligence into military systems have pushed the probability of nuclear conflict to twice its late-20th-century level, a panel of Nobel Laureates warned on Friday.
Speaking at the 75th Lindau Nobel Laureate Meeting, the scientists stated that the annual probability of a nuclear exchange now sits at roughly 2%, translating to a mathematical “half-life” for modern civilization of just 34 years.
The warnings were delivered by the Nobel Laureate Assembly for the Prevention of Nuclear War, a body of 129 laureates advocating for global disarmament. The panel urged world leaders to establish immediate moratoriums on nuclear testing and legally ban autonomous AI systems from nuclear command chains.
Breakdown of Deterrence
David J. Gross, recipient of the 2004 Nobel Prize in Physics, stated that the bipolar stability of the Cold War—governed by the doctrine of Mutually Assured Destruction (MAD)—has dissolved into an unpredictable, multi-polar arms race.
While international treaties successfully reduced global stockpiles from a Cold War peak of 70,000 warheads, Gross noted that almost all major disarmament frameworks have been terminated, suspended or allowed to expire over the last 50 years.
“We are in the middle of a new arms race,” Gross said, adding that the United States and Russia are no longer bound by bilateral constraints. “China is determined to match them. France and the United Kingdom have announced plans to expand their arsenals.”
Gross argued that game-theoretic models of deterrence fail when applied to three or more competing superpowers, leaving the global security architecture without a stable equilibrium. An additional nine states are currently evaluating whether to develop independent nuclear capabilities.
Silicon Interface
The panel highlighted the intersection of machine learning and strategic weapons systems as an immediate threat to global stability.
W. E. Moerner, winner of the 2014 Nobel Prize in Chemistry, cited recent defense war-gaming simulations in which autonomous AI models consistently escalated conventional conflict scenarios toward nuclear deployment.
“The concept of values does not apply to AI,” Moerner said, questioning how legal or ethical accountability could be enforced in the event of an algorithmic failure.
The Assembly issued a formal policy recommendation requiring that any decision to deploy nuclear weapons maintain “meaningful human control,” mandating that at least two human operators remain directly in the launch loop.
Policy Demands
The laureates detailed a four-part policy framework directed at state governments:
- Test Moratoriums: Universal ratification of the Comprehensive Nuclear-Test-Ban Treaty.
- Bilateral Frameworks: Immediate negotiations between Washington and Moscow to establish a successor to the expired New START Treaty.
- Beijing Inclusion: The mandatory inclusion of China in all future strategic arms limitation talks.
- Human-in-the-Loop: Strict statutory prohibitions against delegating nuclear command decisions to autonomous software.
The group appealed to the 600 young scientists in attendance to leverage their institutional influence to revitalize public disarmament movements, drawing parallels to the scientific activism that secured test-ban treaties in the late 20th century.
“There is not one world leader who says nuclear war is a good thing,” said astrophysicist Brian Schmidt, co-initiator of the assembly. “Everyone actually has the same goal: to reduce the risk.”
The Nobel Laureate Assembly is scheduled to convene at the Vatican in two weeks to discuss AI automation and existential risk with Pope Leo XIV.
