The Pentagon is using AI developed by Project Maven to identify targets for airstrikes

The US military has stepped up its use of artificial intelligence tools since the October 7 Hamas attacks on Israel, according to a new report by Bloomberg. Schuyler Moore, chief technology officer for US Central Command, told the news organization that machine learning algorithms have helped the Pentagon identify targets for more than 85 airstrikes in the Middle East this month.

US bombers and fighter jets carried out these airstrikes against seven facilities in Iraq and Syria on February 2, completely destroying or at least damaging rockets, missiles, drone storage facilities and militia operations centers. The Pentagon had also used artificial intelligence systems to find missile launchers in Yemen and surface fighters in the Red Sea, which it then destroyed in a series of airstrikes that same month.

The machine learning algorithms used to narrow down the targets were developed under Project Maven, Google’s now-defunct partnership with the Pentagon. To be precise, the project involved the use of Google’s artificial intelligence technology by the US military to analyze drone footage and images of flags for further review by humans. This caused an uproar among Google employees: thousands had petitioned the company to end its partnership with the Pentagon, and some even quit altogether because of its involvement. A few months after the employee protest, Google decided not to renew their contract, which ended in 2019.

Moore said Bloomberg that US forces in the Middle East have not stopped experimenting with using algorithms to identify potential targets using drones or satellite imagery even after Google ended its involvement. The military has been testing their use for the past year in digital exercises, she said, but began using targeting algorithms in real-world operations after the Oct. 7 attacks by Hamas. However, she clarified that human workers are constantly checking and verifying the AI ​​systems’ targeted recommendations. It was also the human staff who suggested how to organize the attacks and what weapons to use. “There’s never an algorithm that just works, comes to a conclusion, and then moves on to the next step,” she said. “At the end of every step that involves AI, a person checks in.”