Inside the Pentagon's AI war machine

The Pentagon has poured billions into AI warfare, from drone footage analysis to autonomous targeting. Katrina Manson, author of Project Maven and Bloomberg reporter, joins Ian Bremmer to trace how AI went from a computer experiment to key technology for the Pentagon, and why some risks and moral stakes remain unresolved.

The US military's AI shift started in 2017 with Project Maven, a public-private effort to bring artificial intelligence into Pentagon operations. What began as a narrow program to analyze drone footage has grown into something far larger. Today, AI helps identify targets, speed up strike decisions, and support operations from Iran to Venezuela. Bloomberg reporter Katrina Manson, author of a new book on Project Maven, joins GZERO World to trace how that happened, and what the risks look like from the inside. The tech is already showing real limits and raising ethical questions. Algorithms trained in one environment fail in another. Operators are increasingly clicking "accept" without visibility into what the machine is surfacing or why. And as Manson puts it, "the US military is aware" of AI's susceptibility "to sycophancy, to escalation, to bias and hallucination," but "they do not yet have sufficient fixes."Together, they also get into the Anthropic fallout, the US-China AI race, and what happens when commercial AI ethics collide with military requirements.