What happened
A peer-reviewed RAND research report, commissioned by the UK Foreign, Commonwealth & Development Office, argues that attacks which disrupt, deny, or manipulate AI-enabled military and critical-infrastructure systems are an under-governed strategic-stability risk. Its central finding: 'Governments should treat adversarial AI as a strategic stability risk, not merely a cyber security or AI assurance issue.' The study analyses five scenarios and finds adversarial AI may tip the offence-defence balance, that ambiguity between model failure and deliberate attack creates miscalculation risk, and that entanglement of conventional and nuclear C3 raises escalation risk; it recommends embedding the issue in international forums, allied threat-information sharing, after-action reviews, and robust design of AI-enabled military and NC3 systems.
Why it matters
Reframes adversarial AI from a technical cyber problem into a deterrence and crisis-stability concern — directly relevant to national-security strategy and to any enterprise serving defence or critical-infrastructure customers.
Action needed
Brief security and government-affairs leadership on the strategic-stability framing and align AI-assurance controls with recovery, attribution, and fallback planning.