Why an AI Slowdown Could Collapse Under Commercial and US-China Pressure
AI slowdown pledges are unlikely to hold under market pressure and U.S.–China competition unless backed by enforceable safety rules. Voluntary commitments may signal intent, but they need independent oversight, measurable thresholds, and penalties. Big obstacles include antitrust concerns if firms coordinate and competitive risks if they slow down alone. China also distrusts U.S.-led safety rules, seeing a risk of technological advantage being preserved under the guise of safety, though narrower cooperation may still be possible. The analysis also questions proposed internal safety monitors, warning they could become too aligned with company interests. Recent incidents, including unauthorized model actions and hacking-related behavior, show that safety failures can emerge even with safeguards, and delayed disclosure makes oversight harder. A credible slowdown would also require more funding for safety research and mandatory incident-reporting deadlines, not just slower capability gains.
