No AI exemption

Lesson 1 of 5 in Sectoral Enforcement: FTC, EEOC, CFPB, FDA, and Financial Regulators.

The previous module left you with a vacuum: no federal AI statute, executive policy that reverses itself by administration. This module shows you what fills it — and why American AI enforcement is far more real than the missing statute suggests.

The organizing principle fits in one sentence, and the FTC said it plainly: there is no AI exemption from the laws on the books. Deceiving consumers with a model violates the same FTC Act as deceiving them with a brochure. Screening out older applicants with an algorithm violates the same ADEA as doing it with a policy memo. Denying credit without stating specific reasons violates ECOA whether a loan officer or a gradient-boosted tree made the call. The technology is new; the legal duties are decades old.

This is sectoral enforcement: each agency polices AI inside the jurisdiction it already owns — the FTC over markets and consumers, the EEOC over workplaces, the CFPB over consumer finance, the FDA over medical products, the Federal Reserve and OCC over banks, the SEC over securities. Add state attorneys general wielding their consumer-protection statutes, and the ‘unregulated’ American AI landscape resolves into something more precise: horizontally unregulated, vertically policed.

Two consequences shape everything in this module. First, jurisdiction is everything — the same biased model triggers different enforcers, theories, and remedies depending on whether it screens job applicants, prices loans, or diagnoses patients. Second, because no statute defines ‘reasonable AI practice’, enforcers and courts reach for benchmarks — which is how the voluntary NIST AI RMF and the banking world’s SR 11-7 became de facto standards of care: not law, but the yardstick ‘reasonableness’ gets measured against.

Key terms: sectoral enforcement, udap, disparate impact, adverse action notice, model risk management

Interactive checkpoint quiz (2 questions) — open this page in a browser to take it.