We were the employer.
What we learn running the verified human side of AI training — published, with receipts.
We've been running this supply chain quietly behind some of the leading data-annotation companies. Now we're formalizing it — a verified, employed, compliant workforce for AI training.
Anthropic's distillation disclosure was an input-side identity failure. The same failure mode is sitting, mostly unexamined, on the human side of every RLHF pipeline.
Self-reported titles are marketing. Contracts, payroll, and tenure are evidence. The difference shows up in your training data.
Most AI data vendors run on independent contractors. Misclassification claims are now being litigated — and the exposure lands on whoever's left holding the workforce.
Electricians, CNC operators, aviation techs, agronomists: the expertise frontier models need next doesn't browse gig marketplaces. It's already in our payroll data.
Multilingual data pipelines run on self-rated fluency — the most inflated metric online. Employment history is the fluency signal that can't be gamed.