Ian Bremmer sits with Bloomberg defense tech reporter Katrina Manson, who spent years reporting on Project Maven for her new book on the Pentagon's AI push.

The program launched in 2017 with a narrow mandate: use machine learning to process drone footage. It has since expanded into something far more ambitious. Autonomous weapons, drone swarming technology, and AI-assisted targeting are now central to how the Pentagon talks about modern warfare.

The tech rollout is fast, but not reliable. Algorithms fail when the battlefield changes. The targeting process is accelerating to the point where operators are clicking through AI recommendations with little ability to question them. Manson says the military knows about AI's vulnerability "to sycophancy, to escalation, to bias and hallucination," and has not yet found adequate solutions.

Bremmer and Manson also take on the US-China AI race, the IDF's use of AI in Gaza, and what the Anthropic contract dispute reveals about the fault lines between commercial AI ethics and military requirements.


Subscribe to the GZERO World with Ian Bremmer Podcast on Apple Podcasts, Spotify, or your preferred podcast platform, to receive new episodes as soon as they're published.

More For You

Last week, Microsoft highlighted its Ratepayer Protection Tariff (RPT) proposal, a framework designed to ensure datacenters bear the costs associated with required grid improvements rather than shifting those costs to residential customers. Developed in response to growing energy demand in Nevada, the proposal aims to support grid expansion and reliability while providing a model that could inform broader discussions about how states manage rising electricity demand. Read more here.