US AI Safety Staff Shift Closer to White House, Sparking Independence Fears

A small but crucial US government division tasked with assessing the safety of new artificial intelligence systems is being drawn closer into the White House orbit, raising questions about its independence. According to recent reporting, staff at the federal AI safety center are now effectively reporting into White House structures, even as the office lacks a permanent director and has stopped publishing the evaluation reports it once released to the public.

Critics warn that placing civil‑service evaluators under tighter political control could make it harder for them to challenge powerful AI developers or raise alarms about risky frontier models built by companies such as OpenAI and Google. Supporters, however, argue that proximity to the president could finally bring the tiny, under‑resourced unit the budget, staff, and clout it needs to matter. The shift comes as major AI firms build out their own internal and third‑party testing programs, prompting some experts to ask what unique role government evaluators should play. How Washington resolves that tension—between political accountability, technical independence, and industry self‑regulation—will shape the future of AI oversight in the US.