National Statistics Integrity

[DD+POD] NYT documented official labor statistics systematically undercounting AI's economic effect (surveys built to count jobs gained/lost, not roles quietly reorganized around AI collaboration) as June's report showed sector-level contraction the instruments could not clearly attribute to AI.

[DD+POD] Stanford Digital Economy Lab and California Policy Lab develop new instruments to distinguish AI-attributable job displacement from augmentation in employment data: first dedicated academic measurement infrastructure for AI labor impact; BLS official metrics remain without AI-attribution capability.

[DD+POD] California's Employment Development Department and California Policy Lab launched a real-time AI job-loss tracker linking AI-exposure measures to monthly unemployment-insurance claims: first state statistical agency building continuous public measurement infrastructure for AI-attributed displacement, filling a gap the federal apparatus has not addressed.

[DD+POD] UVA/Anthropic/Bank of Canada economists estimate nominal AI GDP at ~$250B in 2025, growing ~2,600% annually in quality-adjusted terms but near-invisible in official statistics because per-capability price falls nearly as fast as supply rises: "a windfall that cannot be seen cannot be shared."

UK science and energy departments carried AI datacentre forecasts differing by roughly 100x until journalism forced DSIT’s 24-hour correction to 34-123 MtCO2e.

Nature documents national statistics crisis across multiple countries: reliability and independence of official data under systemic pressure globally.