American and Chinese experts call for red lines on nuclear AI

American and Chinese experts are calling for red lines on the use of artificial intelligence in the military domain, particularly in systems linked to nuclear weapons, warning that a technical error or misinterpretation of a cyberattack could lead to serious military escalation. Concerns center on the possibility of AI systems being used to monitor cyberattacks or handle nuclear command and control networks, which could shorten the time available for decision-makers.

American and Chinese experts are calling for "red lines" on the use of artificial intelligence in the military domain, particularly in systems linked to nuclear weapons. The experts warn that a technical error or misinterpretation of a cyberattack could lead to serious military escalation. Concerns center on the possibility of AI systems being used to monitor cyberattacks or handle nuclear command and control networks, which could shorten the time available for decision-makers to verify the nature of an attack before taking military action. Among the prominent proposals is keeping decisions with major consequences, including the use of nuclear weapons or launching attacks on nuclear command and control systems, under human control, and preventing intelligent systems from making such decisions independently. Experts also proposed creating a dedicated US-China military communication channel for AI-related incidents, to enable both sides to communicate quickly when ambiguous or unexpected cyber activity occurs. These calls come amid the accelerating integration of AI into military domains and growing concerns about the impact of automated system speed on decision-making time. Experts noted that a new communication channel alone would not be sufficient to prevent crises unless accompanied by clear rules. These proposals remain within the framework of expert discussion, and neither the United States nor China has announced a new agreement on them.