EEOC Guidance on AI Hiring Tools and the ADA
Original Air Date: September 21, 2022
The EEOC removed its 2022 technical assistance on artificial intelligence and the Americans with Disabilities Act from its website in early 2025, following Executive Order 14179. The ADA itself is unchanged.
An employer whose hiring tool screens out a qualified applicant with a disability faces the same liability today that it faced in 2022. What has gone is the federal roadmap for avoiding it. The analysis in this session comes from the statute rather than the guidance, so it still holds.
What Employers Need to Know
-
Title I of the ADA prohibits disability discrimination by employers with 15 or more employees. That threshold and that prohibition are untouched by the guidance removal.
-
A screen out happens when a disability prevents an applicant from meeting a selection criterion, or lowers their score on it, and they lose the opportunity as a result.
-
An algorithmic tool violates the ADA if it screens out an applicant who could do the job with a reasonable accommodation, whether or not the employer knew it would.
-
A vendor's tool is the employer's problem. Liability follows the employer who deploys the tool, not the company that built it.
-
Tools that infer an ability from a correlated trait carry more risk than tools that measure the ability directly. Rating report-writing skill by personality similarity to successful incumbents is the example worth remembering.
-
Announcing that accommodations are available, and how to request them, has to happen before the assessment rather than after a candidate has already been scored.
Frequently Asked Questions
-
Is the EEOC's 2022 AI and ADA guidance still in effect?
The document is no longer published. EEOC removed its May 2022 technical assistance on the ADA and AI, and its May 2023 Title VII guidance on algorithmic tools, after the rescission of the 2023 AI executive order and Executive Order 14179 in January 2025. The ADA and Title VII themselves were not changed.
-
What does screen out mean under the ADA?
Screen out is when a disability keeps an applicant from meeting a selection criterion, or lowers their score on it, and they lose the opportunity as a result. It can happen by reducing the accuracy of the assessment, by creating circumstances the tool does not account for, or by preventing the person from taking it at all. The same analysis drives Section 503 obligations on disability and selection.
-
What are examples of AI tools screening out applicants with disabilities?
Four from this session. A chatbot programmed to reject applicants with employment gaps rejects someone whose gap was medical treatment. Video interview software analyzing speech patterns misreads an applicant with a speech impediment. A gamified memory test requiring a 90 percent score excludes an applicant with a vision impairment who has an excellent memory. A personality assessment screens out an applicant with major depressive disorder on a question about optimism.
-
Can an AI hiring assessment count as a medical examination?
It can. The factors are whether a health care professional administers or interprets it, whether it is designed to reveal an impairment, whether it is invasive, whether it measures physiological responses during a task, and whether it is normally given in a medical setting. Tools that meet that bar carry additional ADA restrictions.
-
What should an employer ask an AI hiring vendor?
Three questions from the session. Was the applicant-facing interface built to be accessible? Are materials offered in alternative formats, and which ones? Did the vendor test whether the algorithm disadvantages people with disabilities, including whether any trait it measures correlates with a disability? A vendor who cannot answer the third has not done the work.
-
Does the guidance removal reduce an employer's ADA risk?
No. Guidance describes how an agency reads a statute. It does not create the obligation, and withdrawing it does not remove one. Employers now have less published federal direction while state AI employment laws in California, Texas and Connecticut come into force. The compliance question got harder.
Speakers
Angie Rosenbaum, Ph.D., Principal Consultant
Kermit Olson, Ph.D., Associate Principal Consultant