The IMF's inequality warning on AI is better than it sounds, and the US regulatory picture is messier than it looks

IMF chief Kristalina Georgieva is now arguing explicitly that AI will worsen inequality in most scenarios unless governments intervene, pointing to her institution's own estimate that roughly 60% of jobs in advanced economies face material exposure, per IMF analysis. The globalization parallel she draws is the serious part: in the 1990s and 2000s, aggregate GDP gains masked concentrated regional losses, and the political fallout took two decades to fully register. Meanwhile, the US regulatory posture is fracturing along two fault lines simultaneously. The White House has imposed opaque, case-by-case controls on Anthropic without clear statutory authority, which Congress is now using as a pretext to revive comprehensive AI legislation it could not agree on a year ago. Separately, the DOJ has framed xAI's unpermitted gas turbines as a matter of national and energy security, which signals that large-scale AI compute infrastructure will now attract security-agency scrutiny alongside environmental review. For founders and investors building AI infrastructure or deploying frontier models with US federal exposure, the relevant shift is that the regulatory risk has moved from the FTC and FCC into the DOJ and executive branch, where the rules are less predictable and the levers more blunt.