Employment · Discrimination
Can an AI Hiring Tool Discriminate Against You? What New York Law Says
Employers increasingly let software decide who gets an interview. The software is not exempt from the discrimination laws, and in New York the rules are stronger than the federal ones.
If you have applied for a job recently, there is a good chance a computer read your application before a person did. Résumé screeners, chatbot interviews, recorded video assessments and online “fit” tests now score, rank and reject candidates at a scale no human recruiter could match. A rejection that arrives an hour after you apply, at two in the morning, was probably not a person’s decision.
Is that legal? The short answer: using AI to hire is legal. Using AI that screens people out because of their race, sex, age, disability or another protected trait is not, and “the software did it” is not a defense.
How a neutral-looking tool can discriminate
Most AI hiring tools are not programmed to reject older workers or women. The problem is how they learn. A tool trained on the résumés of a company’s past “successful” hires learns to prefer whatever those people had in common. If the past hires were mostly young men, the tool can learn to favor signals that track youth or sex: graduation years, gaps in employment, certain schools, sports, even word choice. A timed online test can screen out a candidate whose disability affects speed but not ability to do the job. A video tool that scores facial expressions or speech patterns can penalize an accent, a stutter or a neurological condition.
None of this requires anyone to intend discrimination. The law has dealt with that problem for more than fifty years.
The federal rule: disparate impact
In Griggs v. Duke Power Co., 401 U.S. 424 (1971), the Supreme Court held that Title VII “proscribes not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” Id. at 431. “[G]ood intent or absence of discriminatory intent does not redeem employment procedures or testing mechanisms that operate as ‘built-in headwinds’ for minority groups and are unrelated to measuring job capability.” Id. at 432.
Congress wrote that rule into the statute in 1991. A hiring practice that causes a disparate impact because of race, color, religion, sex or national origin is unlawful unless the employer proves it is “job related for the position in question and consistent with business necessity,” and even then the applicant can win by showing a less discriminatory alternative the employer refused to adopt. 42 U.S.C. § 2000e-2(k)(1)(A).
One provision matters a great deal for AI. Ordinarily, the applicant must identify the particular practice that causes the disparity. But if “the elements of a respondent’s decisionmaking process are not capable of separation for analysis, the decisionmaking process may be analyzed as one employment practice.” 42 U.S.C. § 2000e-2(k)(1)(B)(i). A machine-learning model that weighs thousands of variables in ways no one can explain is a strong candidate for that rule.
The disability law is even more direct. The Americans with Disabilities Act prohibits “selection criteria that screen out or tend to screen out an individual with a disability or a class of individuals with disabilities” unless they are job related and consistent with business necessity, and it requires an employer to give tests in a way that measures the skill being tested rather than an applicant’s impaired sensory, manual or speaking skills. 42 U.S.C. § 12112(b)(6), (7). If you have a disability and are asked to take an automated assessment, you can ask for an accommodation. See how to ask for a reasonable accommodation.
Age is covered too, but more narrowly. The Supreme Court held in Smith v. City of Jackson, 544 U.S. 228 (2005), that the federal age discrimination law allows disparate impact claims, but “the scope of disparate-impact liability under ADEA is narrower than under Title VII.” Id. at 240. An employer has a defense if the practice rests on a reasonable factor other than age. 29 U.S.C. § 623(f)(1).
Can the software company be sued?
Usually the claim is against the employer, which is responsible for the tools it chooses. But the most closely watched case in this area tests whether the vendor can be liable too.
In Mobley v. Workday, Inc., a job applicant over 40 alleged he was rejected from more than 100 jobs at employers that used Workday’s screening tools, once less than an hour after he applied. The court held that Workday could be liable as the employers’ “agent,” because the employers had allegedly delegated to its software the traditional hiring function of rejecting candidates or advancing them. 740 F. Supp. 3d 796, 805–07 (N.D. Cal. 2024). Ruling otherwise, the court wrote, “would allow companies to escape liability for hiring decisions by saying that function has been handed over to someone else (or here, artificial intelligence).” Id. at 806. The court later provisionally certified the age claims as a collective action, and thousands of applicants have opted in. Mobley v. Workday, Inc., 2026 WL 1906986, at *1 (N.D. Cal. July 2, 2026). The case is still in the district court, and no appeals court has ruled on the agent theory.
An open question for older applicants
There is a real split on whether job applicants, as opposed to current employees, can bring an age disparate impact claim under federal law at all. Two federal appeals courts, sitting en banc, have said no, reading the statute’s reference to an individual’s “status as an employee” to exclude outside applicants. Kleber v. CareFusion Corp., 914 F.3d 480 (7th Cir. 2019) (en banc); Villarreal v. R.J. Reynolds Tobacco Co., 839 F.3d 958 (11th Cir. 2016) (en banc). The court in Mobley disagreed and held that applicants are covered, 740 F. Supp. 3d at 811 n.4, and in July 2026 it refused to send the question to the appeals court before trial, while acknowledging there is “substantial ground for difference of opinion.” 2026 WL 1906986, at *1. The Second Circuit, which covers New York, has not decided it.
The better reading protects applicants. Hiring is where age screening does its damage: an older worker screened out at the front door never becomes an employee who can sue. A reading that protects people only after they are hired leaves the most common form of age-biased screening beyond the statute’s reach.
For New Yorkers, though, the federal split matters less than it seems, because State and City law cover applicants either way.
New York law is stronger
New York State. Since December 19, 2025, the State Human Rights Law expressly provides that, in any employment discrimination case, “an unlawful discriminatory practice may be established by a practice’s discriminatory effect, even if such practice was not motivated by a discriminatory intent.” Exec. Law § 296(5-b)(a). Once you prove the practice “caused or predictably will cause” a disparate impact on a protected group, the employer must prove it is job related and consistent with business necessity, and you can still win by showing a less discriminatory alternative. Exec. Law § 296(5-b)(d). That covers every class the State law protects, including age, and it applies to employers across the state, not just in New York City.
New York City. The City Human Rights Law has its own disparate impact provision, and it is especially well suited to opaque technology. If you show that “a group of policies or practices results in a disparate impact,” you “shall not be required to demonstrate which specific policies or practices within the group results in such disparate impact.” N.Y.C. Admin. Code § 8-107(17)(a)(2). The employer must then prove that each practice bears “a significant relationship to a significant business objective” or does not contribute to the disparity. Id. See the guide to the NYC Human Rights Law.
New York City’s AI hiring law
New York City also has a law aimed directly at these tools, Local Law 144. An employer or employment agency may not use an “automated employment decision tool” to screen candidates for jobs or promotions in the City unless the tool has had an independent bias audit within the past year and a summary of the results is posted on the employer’s website. N.Y.C. Admin. Code § 20-871(a). Candidates who live in the City must be told, at least ten business days in advance, that a tool will be used and what qualifications and characteristics it will assess, and the notice must allow them to request an alternative selection process or accommodation. N.Y.C. Admin. Code § 20-871(b)(1)–(2). You can also make a written request for the type and source of data the tool collects and the employer’s data retention policy, and the employer must respond within 30 days. N.Y.C. Admin. Code § 20-871(b)(3).
The law has real limits. The required audit tests only for impact by sex and race or ethnicity (the categories employers report to the federal government), not age or disability. N.Y.C. Admin. Code § 20-870. Enforcement belongs to the City, not to applicants. N.Y.C. Admin. Code § 20-873. And a December 2025 audit by the State Comptroller found the City’s enforcement weak: the agency received only two complaints in two years, and where it reviewed 32 companies and found one possible violation, the auditors found at least 17.
But the law does not displace your other rights. It “shall not be construed to limit any right of any candidate or employee … to bring a civil action.” N.Y.C. Admin. Code § 20-874. And the audit summaries an employer must publish are a useful source of evidence: they report the selection rates the tool produces for different groups.
What has changed in Washington, and what has not
In January 2025 the EEOC took down its guidance on AI in hiring, and in April 2025 an executive order directed federal agencies to deprioritize disparate impact enforcement. Exec. Order No. 14281, 90 Fed. Reg. 17537 (Apr. 28, 2025). That changes what the federal agencies will pursue. It does not change the statutes or Griggs. Individuals can still bring these claims in court, and New York’s State and City laws are enforced independently of any federal policy.
If you think a hiring tool screened you out
- Save everything. The job posting, your application, any assessment invitations, and every rejection email with its date and time stamp.
- Look for the notice. If you live in New York City, check whether you were told a tool would be used. Search the employer’s website for its bias audit summary.
- Make the written request. In the City, ask in writing for the data the tool collects and the employer’s retention policy.
- Ask for an accommodation before taking an automated assessment if a disability could affect your results.
- Notice patterns. Instant or off-hours rejections from jobs you were plainly qualified for, across many employers using the same platform, are what made Mobley plausible.
- Watch the deadlines. A federal charge must be filed with the EEOC within 300 days in New York. State and City claims have longer deadlines, but waiting costs evidence. See New York employment claim deadlines.
AI hiring tools are new. The rule that a hiring practice cannot shut people out because of who they are is not. If you were screened out and want to know whether you have a claim, I can review it with you, or you can start with the case evaluator.
This article is general information about New York and federal law, not legal advice, and it does not create an attorney-client relationship. Attorney Advertising. Prior results do not guarantee a similar outcome.