Skip to content
Compliance7 min read

AI Hiring Laws in 2026

A plain-English guide to AI hiring rules in 2026: NYC bias audits, the EU AI Act's 2027 deadline, Colorado, the UK, Ontario, India and Singapore.

By the HireRabbit.AI team · Published · Last updated

Hiring documents flow through a circuit network toward balanced scales of justice under a ring of EU stars
On this page
  1. What most of these laws ask for
  2. NYC Local Law 144: bias audits and notice
  3. The EU AI Act: high-risk from December 2, 2027
  4. Colorado SB 26-189: explanations and human review
  5. Illinois: video interviews and HB 3773
  6. California's automated-decision rules
  7. Federal law and Mobley v. Workday
  8. United Kingdom: the Data (Use and Access) Act
  9. Canada: Ontario and Quebec
  10. India: the DPDP Act
  11. Singapore: fair hiring now, the Workplace Fairness Act next
  12. Texas, Brazil and Australia
  13. An employer checklist
  14. How HireRabbit.AI handles these duties

This article is general information, not legal advice. Talk to employment counsel about your own situation before you change a process because of anything here. Several deadlines in this area moved during 2026.

If your team uses software to screen resumes, rank applicants or run first-round interviews, the rules that apply to you got much more concrete this year. The EU set a fixed date for its high-risk rules, Colorado replaced its first AI law with a narrower one, and a lawsuit against Workday showed that screening vendors can face claims alongside employers.

Outside the US and EU, the UK rewrote its rules on automated decisions in February 2026, Ontario began requiring AI disclosure in job postings in January 2026, and India's data protection law reaches full force in May 2027.

Ten posts on the rail mark dates from this guide, from NYC in July 2023 to the EU AI Act on December 2, 2027. Move along the rail to raise the dates near you and read each one.

What most of these laws ask for

The statutes read differently, but they keep coming back to three duties. Notice: tell people an automated tool is scoring them. Explanation: someone on your side should be able to say how a tool shaped a decision, especially a rejection. Human review: a person who is really involved has the power to overturn what the tool produced.

NYC Local Law 144: bias audits and notice

New York City has enforced its law on automated employment decision tools since July 5, 2023. It covers automated tools that substantially assist in screening candidates or employees for jobs in the city. Before you use a covered tool, you need:

  1. An independent bias audit within one year before use, reporting impact ratios across sex and race/ethnicity categories and their intersections.
  2. A public summary of the audit results, posted before use.
  3. Notice to candidates at least 10 business days before use, saying which qualifications the tool assesses and how to request an alternative process or an accommodation.
Three weeks of a calendar, with the tool first used on the tall post, a Friday. Weekends do not count, so the notice has to go out by the Friday two weeks before; point at a day to send notice then and count the business days left.

Reported penalties run up to $1,500 per violation per day. A December 2025 New York State Comptroller audit found the city's enforcement ineffective, so expect more of it.

The audit is the employer's duty, even when a vendor supplies one, so check its date against the one-year window.

The EU AI Act: high-risk from December 2, 2027

Annex III of the EU AI Act lists recruitment tools that filter applications and evaluate candidates as high-risk. That brings duties on logging, transparency, human oversight and documentation, split between providers and the employers who deploy them.

The EU's Digital Omnibus on AI, adopted in mid-2026, moved these obligations from August 2, 2026 to December 2, 2027. If you hire into the EU, or anyone uses your AI system's output there, assume the Act matters to you.

Since February 2, 2025, Article 5(1)(f) has prohibited AI systems that infer the emotions of a person in the workplace, including in recruitment, with narrow medical and safety exceptions. Fines for prohibited practices can reach €35 million or 7% of worldwide turnover.

Colorado SB 26-189: explanations and human review

Colorado repealed its 2024 AI law, SB 24-205, on May 14, 2026 and replaced it with the narrower SB 26-189, which takes effect on January 1, 2027. Deployers of automated decision systems in employment decisions must:

  1. Give notice that they use automated decision-making.
  2. Explain an adverse decision in an easily understandable way within 30 days.
  3. Let people access and correct the personal data the system used.
  4. Offer meaningful human review and reconsideration where feasible.

The Attorney General enforces the law, and there is no private right of action. Proposed rules (4 CCR 904-6) were filed on August 11, 2026 and heard on October 26. As of November 2026 there are no final rules, and the Attorney General has said he will not enforce SB 26-189 until rulemaking concludes.

Illinois: video interviews and HB 3773

Since January 1, 2020, the Artificial Intelligence Video Interview Act (820 ILCS 42) has applied when an employer uses AI to analyze recorded video interviews for Illinois-based positions. Before the interview, you must tell the applicant AI may be used, explain how it works and what general types of characteristics it evaluates, and get consent. You must delete an applicant's videos, including copies, within 30 days of a request.

HB 3773, in force since January 1, 2026, amended the Illinois Human Rights Act to cover AI discrimination in employment decisions and requires notice when AI is used. Detailed notice rules were proposed and withdrawn in 2026.

California's automated-decision rules

California's Civil Rights Council rules, effective October 1, 2025, confirm that anti-discrimination law covers automated decision systems. They apply to employers with five or more employees and require them to keep automated decision system data for four years.

Under the CCPA, businesses already using automated decisionmaking technology (ADMT) for significant decisions such as hiring must comply with the ADMT rules by January 1, 2027. AB 1883, signed September 30, 2026, bans AI tools that infer how employees feel from January 1, 2027, though it is unclear whether it reaches applicants.

Federal law and Mobley v. Workday

Mobley v. Workday alleges that Workday's screening tools led to the plaintiff's rejection from more than 100 jobs because of his age, race and disability. In July 2024 the court let disparate-impact claims proceed on the theory that a vendor can act as an employer's "agent." On May 16, 2025 it conditionally certified a nationwide collective of applicants aged 40 and over.

More than 15,000 people have opted in, and on September 14, 2026 the plaintiffs moved for Rule 23 class certification. Our Mobley v. Workday explainer covers the full case. The lesson so far: "the vendor's tool did it" does not protect the employer.

Federal enforcement moved the other way. The EEOC removed its AI guidance in January 2025, and in October 2026 it sent the White House's regulatory review office a final rule rescinding the interpretive parts of the Uniform Guidelines. Title VII, the ADA and the ADEA still apply, and private plaintiffs can still sue.

United Kingdom: the Data (Use and Access) Act

Since February 5, 2026, the Data (Use and Access) Act 2025 has replaced UK GDPR Article 22 with Articles 22A to 22D. A solely automated significant decision, such as auto-rejecting an applicant, can now rest on any lawful basis unless it involves special category data such as health or ethnicity. Whatever the basis, the candidate must be told and be able to make representations, get human intervention and contest the outcome.

The ICO's March 2026 Recruitment Rewired report found many employers making solely automated decisions without realising it.

Canada: Ontario and Quebec

Since January 1, 2026, Ontario employers with 25 or more employees must disclose in a publicly advertised job posting if they use AI to screen, assess or select applicants. They must tell interviewed applicants the outcome within 45 days of the last interview and keep postings and those notices for three years.

Since September 22, 2023, Quebec's Law 25 has required any enterprise making a decision based exclusively on automated processing to tell the person. On request, it must explain the information used and the principal factors, and let the person submit observations to staff who can review the decision.

India: the DPDP Act

India's DPDP Rules were notified in November 2025, and the main duties, including notice and consent, security safeguards and breach reporting, apply from May 2027. Resumes submitted online are covered. Where data drives a decision about a person, it must be complete, accurate and consistent, and it must be erased once it has served its purpose. Penalties run up to ₹250 crore.

Singapore: fair hiring now, the Workplace Fairness Act next

Singapore has no AI-specific hiring law. The Workplace Fairness Act, passed in January 2025, covers hiring decisions and is expected to take effect by the end of 2027. Until then the Tripartite Guidelines on Fair Employment Practices apply, and TAFEP's October 2025 advice says to use only job-related data, let people make final decisions and tell candidates how you use AI.

Texas, Brazil and Australia

Texas's TRAIGA took effect on January 1, 2026, targets intentional discrimination and is enforced only by the Attorney General. Brazil's LGPD lets people ask for review of solely automated decisions. In Australia, from December 10, 2026, Privacy Act privacy policies must describe personal information used in significant automated decisions.

An employer checklist

Treat this as a starting point for a conversation with counsel.

Each padlock is one of the ten places in step 2 below, from NYC to Singapore. Point at a lock to open it and see the main duty that place asks of you.
  1. Inventory every tool that affects who moves forward.
  2. Map your jurisdictions. NYC, Colorado, Illinois, California, the EU, the UK, Ontario, Quebec, India and Singapore each have different triggers.
  3. Commission bias audits where required, such as NYC, and keep the results.
  4. Send candidate notices early enough, which in NYC means at least 10 business days before use, and add the AI statement to Ontario job ads.
  5. Make every automated score explainable and keep a person in charge of rejections.
  6. Give candidates a route to challenge an automated decision.
  7. Retain records for the longest period that applies, such as four years in California.
  8. Question vendor claims about detecting emotion or personality from video or voice.

How HireRabbit.AI handles these duties

HireRabbit.AI scores each resume on skill match, work experience, projects and education, with weights shown to the recruiter.

Auto-reject runs only on a recruiter-set cut-off and a confirmed batch, and you can undo it. Candidates see a consent screen before an AI interview. HireRabbit.AI has not yet completed an independent bias audit, so NYC employers would need to plan for one.

Put this into practice.

Free during invite-only early access. All you need is a work email, and a person reviews every request.