Pentagon technology chief Emil Michael, a task force vice chair responsible for coordination with AI companies, said recent model incidents stemmed from inadequate sandboxes and log checks. “Everything we’ve seen so far was preventable by the labs themselves,” he said. Task force leader Jay Clayton argued on CNBC that slowing AI development for safety gets it backwards and said the government is in dialogue with 20 to 25 AI companies.
The Commerce Department has no representative on the task force, despite its role in AI testing. Its CAISSI office tests and evaluates models and holds agreements with AI companies governing how those evaluations work. The task force has 120 days to report on AI’s risks, opportunities and the federal government’s responsibilities.