Remove zip codes from your hiring AI and you fail the EU's data-representativeness test. Keep them and you violate Illinois. A company hiring in New York, Chicago, Denver, Austin and London now falls under six overlapping AI-hiring laws — and they don't agree with each other. Illinois (live January 1) bans zip codes as proxies for protected classes. The EU AI Act, whose high-risk recruitment rules hit August 2, demands training data so representative that geography is hard to remove. NYC requires intersectional race-by-sex impact ratios. Texas rejects disparate-impact claims entirely. One bias audit cannot satisfy all of them — each regime needs a differently-shaped deliverable. And the regulator learned to look. In December a NY State Comptroller audit found 17 LL144 violations in the same 32-company sample where the city had found one — its own reviewers, by their own admission, lacked the expertise to spot them — and the city has agreed to stop waiting for complaints. Of 391 NYC employers studied, only 4.6% had published the bias audit the law requires. The honest answer most vendors won't give: in a few places, full compliance everywhere is mathematically impossible, and leadership has to make a deliberate legal call. Pretending the conflict isn't there is how a 2024 "our tool isn't a regulated hiring tool" memo becomes a 2026 discovery exhibit. Which two regimes in your stack conflict the hardest — and how did you decide which one to break? #AIGovernance #HRTech