The EEOC reset its priorities. Your AI hiring risk did not go away.

What the EEOC actually did

In mid June 2026 the EEOC retired its existing Strategic Enforcement Plan and replaced it with a new National Enforcement Plan covering fiscal years 2025 through 2029. SHRM, Jackson Lewis, Ogletree, and HR Dive all covered the change, which is the largest federal employment enforcement reset of the year.

The headline shifts are in priority, not in law. The new plan puts more weight on DEI programs and on what it frames as reverse discrimination claims, and it scales back some EEO reporting obligations. In plain terms, the agency is signaling where it intends to spend its limited investigative attention for the next several years.

That signal matters for planning. It does not rewrite Title VII, and it does not erase the legal theory that has always governed automated hiring screens. The plan is a statement of where the agency will look first. It is not a statement of what is suddenly legal.

Why your AI hiring risk did not move with it

Here is the part that gets lost in the coverage. The core risk in an AI hiring tool is disparate impact. If a resume screener, a video interview scorer, or a ranking model pushes out one protected group at a measurably higher rate than another, you have exposure, regardless of which claims the EEOC put at the top of its list this cycle.

Disparate impact is a settled legal theory. It existed before this plan and it will exist after the next one. An enforcement plan changes the odds that a federal investigator knocks on your door for a given issue. It does not change whether your tool is producing a discriminatory result on the people moving through it right now.

The other actor the headlines understate is the private plaintiff. Applicants and their lawyers do not need the EEOC to go first. A rejected candidate who suspects an algorithm screened them out can bring a claim on their own. A priority shift at the agency does nothing to slow that path, and in some readings a quieter federal posture invites more private action, not less.

So the honest read is this. The enforcement lens moved. The thing it was pointed at, your screening tools and the outcomes they produce, is still sitting on your desk.

What HR should re-audit this quarter

This is a good moment to run a real audit rather than a checkbox one. The work is specific. A short list of what to actually look at:

• Every automated tool in the funnel. List the resume parsers, the keyword filters, the assessment scorers, the video and chat tools, and any ranking model in your applicant tracking system. You cannot audit what you have not inventoried.
• Selection rates by group. For each tool, look at pass and advance rates across protected categories. A large gap is the early warning. The classic threshold is the four fifths rule, where one group advancing at less than eighty percent of the top group's rate is a flag worth investigating.
• The vendor's own validation. Ask each vendor for their bias testing, their methodology, and the date of their last study. A vendor who cannot produce this is a risk you are carrying on their behalf.
• The human override path. Confirm a person can review and reverse a tool's decision, and that the reversals are logged. A fully automated reject with no human check is the hardest pattern to defend.
• Your contract terms. Check who is liable when the tool discriminates. Many vendor agreements quietly leave that with you.

If you have not done this since the tools went in, the priority shift is the prompt to do it now, while it is routine rather than a response to a complaint.

The documentation that survives a priority shift

The single most useful thing you can build is a paper trail that does not depend on who is running the agency. Political priorities swing. The questions a judge or an investigator asks do not, and they are answerable in advance.

Keep records that show three things. First, that you knew what each tool did and tested it for adverse impact. Second, that the tool measures something the job actually requires, so you can defend it as job related and consistent with business necessity. Third, that a human stayed in the loop and could override a result.

Those records are the same under a plan focused on disparate impact and a plan focused on reverse discrimination. They are the same under the next administration and the one after that. Documentation built to satisfy whoever happens to be in charge this year is fragile. Documentation built around the underlying law holds in either direction.

This federal picture is distinct from the patchwork of state rules. Several states have their own AI hiring requirements, and Colorado in particular has put specific duties on employers. We cover that separately in what HR owes now on AI hiring law. This briefing stays on the federal enforcement reset. The two work together, and you need both in view.

What to do this week

You do not need a task force. Pull your tool inventory and your last set of selection rate numbers, or generate them if they do not exist. Spend an hour finding the one tool with the widest outcome gap and the thinnest documentation. That is your highest exposure, and it is usually obvious once you look.

Then ask each vendor for current bias testing in writing. The request itself is useful, because a vendor who stalls is telling you something. Save what comes back. If you also operate under a state law like Colorado's, line this federal audit up next to that obligation so you are not running two disconnected processes.

None of this requires a verdict on the politics of the new plan. It requires treating your hiring tools as the regulated decision making systems they are, on any given Tuesday, under any given administration.

The skill under the policy

Every cycle brings a new enforcement plan, a new state law, or a new vendor feature, and HR is expected to absorb all of it without dropping the day job. The professionals who stay calm through these resets are not the ones who memorized this year's priorities. They are the ones with a method: a way to audit a tool, document a decision, and keep a human accountable for what an algorithm produced.

That method is what holds when the plan changes again, and it will. If you want the structured version built for people who own hiring and compliance, The Leveraged HR Professional teaches it from the ground up, and the two minute course quiz will point you to the right starting place for your role.