Pentagon's AI weapons and projects often bypass internal guidelines, congressional report finds
A congressional watchdog found that Pentagon officials are building AI-capable weapons and projects that frequently don't comply with the agency's own established AI guidelines.

What happened
A congressional watchdog report published in October documented that the Pentagon is developing weapons and projects with AI capabilities that officials say don't always closely adhere to the Pentagon's internal guidelines for responsible AI use.
Context
This compliance gap suggests a tension between the Pentagon's stated commitment to responsible AI development and actual implementation on the ground. The finding raises questions about enforcement mechanisms and whether guidelines are being treated as binding standards or aspirational principles across the department's AI initiatives.