The Pentagon has secretly approved an updated doctrine for using AI to select targets on the battlefield, according to journalists from Bloomberg. The document, signed in April, is not classified but has not been made public.
The new wording outlines "combat systems in which artificial intelligence initiates actions under human monitoring."
According to the document, the speed of warfare and advancements by adversaries in AI may necessitate the creation of "fully autonomous systems" by U.S. military forces. It claims that the new technology should reduce the cycle "from sensor to target engagement" and increase the pace of operations.
The doctrine expands AI's role in processing intelligence data. Specifically, systems are expected to more quickly correlate information about targets from multiple platforms, creating a more comprehensive picture of the situation. Commanders are encouraged to utilize neural networks more actively for analysis, decision-making, and risk management.
The document also includes provisions for "reducing harm to civilians."
However, the Pentagon warns of the consequences of excessive reliance on algorithms. The text discusses "serious moral and legal dilemmas" and the need for clear ethical guidelines for decisions made by AI. An appendix on the integration of automation emphasizes that it does not replace human thinking and proactive communication.
The agency described the revision of the doctrine as the result of years of work, taking into account exercises, war games, and assessments of the future security environment.
"The Department of Defense ensures that humans are always aware of important operational decisions. The department's AI technologies do not allow for autonomous target selection or strikes. This guarantees that commanders remain accountable for every decision, enabling them to act based on the most current and accurate operational picture," a Pentagon representative stated in comments to journalists.
On the same day as Bloomberg's article was published, the department released a separate press release about the "launch of a network of agents to transform battle management and target selection systems."
"We are creating a compatible network of AI agents that provides commanders with faster access to higher-quality information while keeping human judgment at the center of every target selection decision," noted Cameron Stanley, the Pentagon's chief digital technology officer.
Journalists highlighted that back in 2023, UN Secretary-General António Guterres called for a ban on lethal autonomous weapon systems, labeling them as "politically unacceptable" and "morally repugnant."
It is worth noting that in June, the Pentagon reported a 1775% increase in AI adoption among its personnel.