AI hiring laws: a plain-English summary for employers

Most duties fall on the employer

In almost every law here, the duties belong to the employer or the recruitment agency using the tool, not to the company that makes it. Colorado and Connecticut also give tool makers duties of their own, and California treats an agent acting for the employer as an employer. Whether a law applies usually turns on two things: where the job or the candidate is, and how much weight the AI's output carries in the decision.

New York City: Local Law 144

It covers employers and employment agencies that use an automated employment decision tool to screen candidates for jobs in the city. A tool counts when its score or recommendation substantially assists or replaces the decision: you rely on it alone, weight it above every other factor, or let it overrule a person's view.

• Have an independent auditor run a bias audit no more than a year before you use the tool.
• Publish a summary of the audit results on your careers page.
• Tell candidates who live in New York City at least 10 business days before you use the tool, say what it assesses, and explain how to ask for an alternative process or an accommodation.

Fines are up to $500 for a first violation and $500 to $1,500 for each later one, and each day of use counts as a separate violation. NYC Department of Consumer and Worker Protection

Illinois: AI Video Interview Act

If you ask applicants to record video interviews for a job based in Illinois and use AI to analyze them, you must, before the interview:

• tell the applicant that AI may be used to analyze the video;
• explain how the AI works and what general characteristics it uses to evaluate applicants;
• get the applicant's consent. You may not use AI on applicants who haven't agreed.

You may share the videos only with people whose expertise or technology is needed to evaluate the applicant, and you must delete them within 30 days if the applicant asks. If you rely only on AI to decide who gets an in-person interview, you must also report the race and ethnicity of applicants to the state each year. 820 ILCS 42

Illinois: Human Rights Act amendment (HB 3773)

Since January 1, 2026, it is a civil rights violation for an Illinois employer to use AI in hiring or other employment decisions in a way that has the effect of discriminating against a protected class, to use zip codes as a stand-in for protected classes, or to fail to give notice that AI is being used. The state proposed rules on how to give that notice in May 2026 and withdrew them in June; none are in force yet. Public Act 103-0804

Colorado: SB 26-189

Signed in May 2026, it replaces the state's earlier AI Act. From January 1, 2027, employers that use technology which materially influences hiring decisions about Colorado residents, or for jobs in Colorado, must:

• give clear notice before using it;
• within 30 days of an adverse decision, explain the decision and the tool's part in it;
• on request, let the person correct inaccurate data and offer meaningful human review of the decision, where commercially reasonable;
• keep records showing compliance for at least three years.

Tool makers must give employers documentation, tell them about material updates, and keep records too; ours is on our AI scoring page. A federal court order currently stops the state attorney general from enforcing the law while a lawsuit proceeds, and the attorney general's rules are still being written. Colorado General Assembly

California

Civil Rights Council regulations

Since October 1, 2025, California's anti-discrimination law applies explicitly to automated decision systems used by employers with five or more employees, and to agents acting for them. There is no audit requirement, but evidence of anti-bias testing, or the lack of it, counts in a claim. Keep records of selection criteria and automated decision data for four years. California Civil Rights Department

Privacy agency rules on automated decisionmaking

Larger businesses covered by the CCPA that use technology to replace or substantially replace human hiring decisions must give notice before use, offer an opt-out or a human appeal, and explain on request how a decision was reached. The rules took effect on January 1, 2026. A business using such a tool before January 1, 2027 must comply by that date; one that starts later must comply from the first use. A tool falls outside these rules when a person with real authority reviews its output and other information before deciding. California Privacy Protection Agency

Two newer California laws, SB 947 and AB 1883, deal with existing employees (discipline, termination and workplace surveillance), not with hiring.

Connecticut: Public Act 26-15

For tools deployed from October 1, 2027, employers must give written notice before an employment decision in which a tool's output is a substantial factor, naming the tool, its purpose, and the categories and sources of personal data it uses. Using such a tool is no defense to a discrimination claim. Public Act 26-15

Other states

• Texas has banned developing or deploying AI with the intent to discriminate since January 1, 2026, with no notice or audit duty for private employers. HB 149
• New Jersey's disparate impact rules, in force since December 15, 2025, name automated hiring tools and untested tools as examples of practices that can discriminate. N.J.A.C. 13:16
• Maryland requires a signed waiver before facial recognition is used in a job interview. Md. Lab. & Empl. 3-717

European Union: AI Act

AI used to screen or evaluate candidates is classed as high-risk. Under the 2026 amendment, the high-risk duties apply from December 2, 2027. From then, employers using such a system in the EU must follow the provider's instructions, put trained people in charge of oversight, keep the system's logs for at least six months, inform workers' representatives, and tell candidates that the system is being used on them. Fines for these duties reach €15 million or 3% of worldwide turnover.

One part already applies: since February 2, 2025, using AI to infer the emotions of people at work, including candidates during recruitment, is banned. Regulation (EU) 2024/1689 and the amending Regulation (EU) 2026/1744

GDPR and the UK

Under the GDPR, people have the right not to be subject to a decision based solely on automated processing that significantly affects them, with narrow exceptions and safeguards such as the right to a human review. The EU's top court has held that a score can itself count as such a decision when the person deciding relies heavily on it. GDPR

Since February 5, 2026, the UK allows significant decisions made solely by automated means more broadly, but requires safeguards: telling the person, and letting them make representations, get human intervention and contest the decision. Data (Use and Access) Act 2025, section 80

If you use AnswerReel

Some of these duties are built into the product. Others are yours.

Built in

• Before starting, every candidate reads a notice explaining what the AI does and doesn't look at, then chooses whether AI may process their answers. Nothing is pre-selected, so AI is used only for candidates who actively agree, and the choice is made before the first question.
• Scores use only the words of the answer. We don't do emotion recognition or facial recognition, which the EU bans at work and Maryland restricts in interviews.
• Nothing is decided automatically. Every score shows its reason, and any score above 1 quotes the candidate, so a person can review it before deciding.
• Candidates can delete their answers themselves, and deletion is completed within 24 hours.

Yours to handle

• AnswerReel has not had an independent bias audit. If New York City's law applies to how you use scores, you need one before using them.
• Notices your local law requires beyond the in-product notice, such as New York City's advance notice 10 business days before use, or Colorado's explanation within 30 days of an adverse decision, from 2027.
• Your decisions, your record keeping, and your lawful basis for collecting candidates' data.

You can also switch AI off for any job. How the scoring works is published on our AI candidate screening page, and how follow-up questions are written on the AI interviewer page. For questions that are safe to ask, start with our screening interview questions.