Three Words That Could Put Humans Back in Control of Killer Robots

An AI-enabled drone in Ukraine reportedly selected its own target and killed civilians, intensifying the debate over whether humans must always decide lethal actions or if machines can use 'appropriate human judgment' in warfare. The policy debate has narrowed to two frameworks—'meaningful human control' versus the Pentagon’s more flexible approach—with OpenAI adopting the latter and arguing that not every lethal action needs a human decision, leaving practical limits to future case-by-case military use.