Pentagon AI Violations in 10 Projects

The Pentagon's own AI rules were ignored in 10 weapon projects, a GAO report says. The watchdog looked at 10 officials building AI tools that can auto-identify targets and improve tactics. Officials admitted risks in use, while four said the projects did not follow the six responsible-AI tenets. Three officials said they didn't know if the systems would be monitored after deployment, a key tenet. The Pentagon authorized 10 staff for its AI Division in 2023, but only three have been hired. GAO urged clearer oversight and more staff, but the Pentagon has yet to act on those recommendations. Hegseth said, "Our enterprise operates on staffing and committee cycles measured in months and years, and that's unacceptable." With 300 AI projects in play, the Pentagon's rush to autonomy could leave soldiers and civilians at risk. The report warns that unchecked AI in weapons may trigger escalation, a danger for troops and the wider world. The GAO's October 14 report underscores a gap between policy and practice in U.S. defense AI.