Source: Bloomberg (paywall)
The Department of Defense has formally revised its targeting doctrine to permit AI systems to initiate actions—not merely recommend them—subject to human monitoring after the fact. This departs from previous protocols that required human approval before engagement. Human operators shift from decision-makers to supervisors of automated systems, compressing response times while distributing accountability in ways existing international law and military ethics frameworks were not designed to address. The revision indicates the Pentagon has resolved its internal debate about autonomous weapons in favor of operational speed over the precautionary restraint that public debate and allies have demanded.