Dossier mode

In 2008 the public discovered after the crash that it had insured the wealthy against their own losses. The AI build-out is arranging the same deal before the fact.

Subsidies signed now, job displacement promised now, surveillance harms landing now, the backstop floated now. Gains up, losses down — pre-positioned.

Communities subsidize the data centers — and insiders are often paid to sell the deals.

Municipalities and states offer large tax abatements, cheap power, and infrastructure to attract AI/data centers that employ few once built and strain grids and water — often alongside benefits to well-placed local figures (consultant contracts, board seats). Subsidy pattern = FACT where reported; the insider-capture framing = SOME SMOKE, to be pinned case-by-case to specific contracts and disclosures.

They are telling us out loud that the point is to eliminate the jobs.

In 2025 Anthropic CEO Dario Amodei warned publicly that AI could eliminate up to half of entry-level white-collar jobs within a few years and spike unemployment — widely reported from his interviews; other executives echoed it. FACT that the statement was made (accuracy of the forecast is a separate open question). The public that subsidizes the build-out is being told, by its beneficiaries, that a purpose is to displace its labor. [COI: Anthropic makes the assistant drafting this site — flagged; graded as an attributed on-record quote regardless.]

The surveillance side already produces false positives — and real arrests.

The ACLU documents multiple wrongful arrests from facial-recognition misidentification, beginning with Robert Williams (Detroit, 2020, arrested in front of his family for a crime he didn't commit) and others, disproportionately Black. The concrete, present-tense cost of a technology sold as safety: the errors land on ordinary people while the contracts and valuations accrue upward.

And the bailout is already being floated — before the losses arrive.

Figures in and around the AI industry have raised a government backstop if the bets fail — the government as insurer of last resort, or an implicit 'too important to fail' status for the largest labs, amid open bubble talk. Graded SOME SMOKE and attributed: a real on-record strand of discussion, not a policy or promise; any specific expectation would be pinned to the exact statement. Same tell as 2008 — arrange for the public to hold the downside before it's due.

How this page is graded.

• FACT: that Amodei's job-loss forecast was made (an attributed quote); the ACLU-documented wrongful arrests; that data-center subsidies are offered.
• SOME SMOKE / attributed: the insider-grift framing on specific subsidy deals; the AI-bailout/backstop expectation (discussion, not policy).
• NOT asserted: that a specific AI bailout is coming; that the job-loss forecasts are proven; that every subsidy is corrupt. COI on the Anthropic quote is disclosed.

Help us fill these lines.

• OpenCase-by-case documentation of specific data-center subsidy deals and the insiders who profited from approving them.
• OpenWhether any AI-industry bailout expectation is ever formalized into policy, and on what terms.
• OpenThe realized labor-market effect of AI deployment versus the executives' forecasts.