AI Regulation Tracker  /  Hiring and employment

The EEOC's First AI Hiring Case Is Still the One That Governs Your Recruiting Software

In August 2023 the EEOC settled its first case built on an algorithmic hiring tool. A federal consent decree required iTutorGroup to pay $365,000 after its tutor-application software automatically rejected more than 200 US applicants for being too old. This is the origin precedent for federal AI-hiring enforcement, and it still binds.

By Anthony Guerriero  ·  The Leveraged Years Regulatory Research  ·  Published July 20, 2026  ·  Last updated July 20, 2026

Let me be upfront about the date, because it matters. This is not a bulletin from this week. The EEOC announced this settlement on September 11, 2023, and the events go back to 2020 hiring. I am covering it now on purpose, because it is the origin point for federal AI-hiring enforcement in the United States, and almost every newer state law, agency plan, and lawsuit in this space is arguing with the precedent this case set. If you want to understand where AI-hiring liability comes from, you start here.

The facts are almost comically clean, which is part of why the case landed the way it did. iTutorGroup, a group of companies that hired US-based tutors to teach English to students in China, ran an online application system. According to the EEOC, that system was programmed to automatically reject female applicants who were 55 or older and male applicants who were 60 or older. More than 200 qualified people in the United States were screened out for no reason other than their age. That is a straightforward violation of the Age Discrimination in Employment Act, which protects workers 40 and over.

How the automatic rejection got caught

The tell was simple. One applicant submitted two applications that were identical except for the date of birth. Using a real, older birth date, the application was rejected. Using a more recent birth date, the same person got an interview. That is the kind of clean before-and-after that turns a suspicion into a case. It also exposes the specific danger of an automated screen. A human recruiter who discriminates leaves an ambiguous trail. Software that applies a fixed age cutoff produces a perfectly consistent, perfectly documentable pattern, and that consistency is exactly what makes it easy to prove.

What the consent decree actually required

This settled as a consent decree, which is the part professionals should not skim past. A consent decree is a court order. It is not a private handshake and it is not agency guidance you can weigh and set aside. The court in the Eastern District of New York entered it, and its terms are enforceable as a judgment. Under it, iTutorGroup agreed to pay $365,000 to the applicants who were automatically rejected. It agreed to adopt a robust new anti-discrimination policy, to provide continuing anti-discrimination training to everyone involved in hiring tutors, and to stop requesting applicants' birth dates. It accepted injunctions against age and sex discrimination in hiring. And it agreed to let the EEOC monitor its compliance for at least five years, longer if it resumes hiring tutors in the United States, including a duty to notify and interview previously rejected applicants if it does.

The EEOC was blunt about why it pursued the matter. As the agency's trial attorney put it, "Prohibitions on age and other types of discrimination do not stop at the border." The acting district director framed the stakes just as plainly: "Everyone loses when employers engage in age discrimination." The location of the students, or the fact that a machine made the call, did not soften the analysis.

The rule this case set, and why it still holds

The operating principle from iTutorGroup is one sentence, and it has not aged a day. If your software makes the hiring decision, you are liable for the discrimination it produces. There is no vendor defense in the decree, no "the algorithm did it" mitigation, no carve-out because the screening was automated rather than manual. The EEOC treated the software exactly as it would treat a manager who threw out applications from anyone over a certain age. Automation was not a shield. If anything, it was an aggravator, because the discrimination was systematic and applied to everyone who came through the door.

That is why this 2023 case is still the reference point in 2026. When you read newer instruments in this area, the New York City bias-audit rule for automated employment decision tools, Illinois and Colorado AI-employment laws, the EEOC's own later enforcement planning and the disparate-impact debates around it, they are all building on the same premise iTutorGroup nailed down first. An automated tool that filters people by a protected characteristic is discrimination, and the employer deploying it answers for the result.

What US employers and counsel should take from it

The practical lesson is not complicated, but it is easy to ignore until an agency shows up. First, you are responsible for what your hiring tools do, whether you built them or bought them. "The vendor configured it" is not in this decree, and it will not be in yours. Second, an automated screen produces evidence. The same consistency that makes a tool efficient makes its bias trivially provable, so a tool that encodes an unlawful cutoff is a documented, repeatable violation waiting to be discovered by a paired-application test exactly like the one that cracked this case. Third, the fixes the EEOC demanded are the fixes to make now, before there is a problem. Know what every screening and scoring tool in your pipeline actually keys on. Stop collecting data points like age or birth date that have no lawful place in the screen. Test your tools for disparate outcomes across protected classes. And keep the records that show a human, not just a model, stands behind the decision.

For employment counsel, the value of iTutorGroup is that it gives you a clean, court-entered precedent to put in front of a client who thinks AI hiring tools are a compliance shortcut. They are not. They are a compliance surface. The EEOC proved it can and will treat an algorithmic screen as the discriminating actor, hold the employer liable, extract real money, and supervise the company for years afterward. That was true in 2023, and nothing since has walked it back.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how the ADEA and any AI-hiring rule applies to your situation with qualified employment counsel in the relevant jurisdiction.