Federal retreat on disparate impact pushes AI hiring bias risk to the states — and the courts
Not legal advice. This article discusses legal and regulatory developments and is generated with AI assistance under human editorial review. It is provided for informational purposes only and does not constitute legal advice. Laws and regulations vary by jurisdiction and are subject to change. You should consult your own qualified legal counsel before taking or refraining from any action based on the content of this article. Read our AI Transparency statement.
The News
A Greenberg Traurig alert published 15 September 2026 maps the evolving litigation landscape for AI-assisted hiring. After the Trump administration's April 2025 executive order aiming to "eliminate the use of disparate-impact liability," the EEOC's internal September 2025 memo directing discharge of disparate impact claims, and the DOJ Office of Legal Counsel's 9 June 2026 opinion that EEOC disparate impact guidelines are unconstitutional, federal enforcement has pulled back. In response, California, Illinois, and New Jersey have codified disparate impact liability under their state anti-discrimination laws, and private plaintiffs' attorneys are pursuing AI hiring cases under traditional discrimination theories. The alert flags that discovery in AI cases may extend to model inputs and outputs, training and validation data, selection and weighting criteria, bias testing, vendor contracts, and records of human review — much of which may sit with the vendor, raising possession and trade-secret issues.
Why It Matters
For TA leaders, the practical shift is that the centre of gravity for AI hiring risk has moved from a single federal regulator to a patchwork of state laws and active litigation. Two live cases matter: Mobley v. Workday, where a federal court allowed discrimination claims to proceed against both the AI vendor and the employers using its screening tools; and Kistler v. Eightfold AI, testing whether AI-generated applicant scores make a vendor subject to federal consumer-reporting laws. Both mean employers can no longer assume liability sits with the vendor. The discovery point is the operational risk: if sued, you may need to produce documentation you do not hold, held by a vendor under a contract that does not guarantee access. Pre-litigation diligence — what your vendor can and will provide — is now a governance question, not just a procurement one.
Taking Action
All action is situated in your corporate context. These are starting points, not prescriptions — choose what fits.
Connect
Who to talk with in your enterprise
Your employment litigation counsel: ask what documentation you would need to defend an AI hiring decision in a disparate impact claim, and what proportion of it you actually hold versus your vendor. Your procurement or vendor management lead: review your ATS and screening vendor contracts for clauses covering litigation support, access to model documentation, and preservation of system version data — the alert makes clear these are now litigation-critical.
Inspect
What to take a closer look at
Run a simple litigation-readiness check on your primary AI screening tool: can you produce, on request, the model version used at a specific time, the scoring data for a specific candidate, any bias testing performed, and a record of human review of the system's recommendation? If any of those four is missing, that is the gap to close before a claim, not after.
Reflect
What to think about more deeply
If the federal government steps back from policing AI hiring bias, is the resulting state patchwork a victory for local democracy — or just a more expensive way to leave employers guessing where the line is?