← All resourcesGuide · Global · 12 min read
AI Hiring Laws by Jurisdiction: The U.S. vs. the Rest of the World
Only one place on earth mandates an independent bias audit of AI hiring tools. Yet anti-discrimination law reaches AI outcomes almost everywhere. What really differs between regions is notices, human review, and whether you may even collect the demographic data a bias test needs.
The short version
- Mandatory audit: only New York City.
- Testing as evidence: California, Connecticut and New Jersey treat bias testing as evidence in a discrimination claim.
- Testing as an expectation: the UK regulator and Singapore’s data regulator expect it.
- Duties follow the candidate: they attach to where the job and the candidate are, not where you are headquartered.
United States: federal
There is no federal statute specific to AI in hiring. Title VII, the Age Discrimination in Employment Act and the Americans with Disabilities Act apply to AI tools exactly as they apply to any other selection procedure, and the Uniform Guidelines’ four-fifths rule remains the standard first test for adverse impact. Federal agencies have stepped back from AI guidance, but private disparate-impact litigation continues. In Mobley v. Workday, a federal court allowed claims to proceed against an AI vendor on the theory that it acted as the employer’s agent.
United States: state and local
JurisdictionWhat applies
New York CityLocal Law 144: independent bias audit within a year before use, public summary, 10-business-day notice. In force.
CaliforniaFEHA automated-decision rules (in force since Oct 1, 2025): discriminatory outcomes are unlawful, anti-bias testing counts as evidence, records kept four years. CCPA automated-decision rules add notice, opt-out and a risk assessment from Jan 1, 2027.
ColoradoSB 26-189 takes effect Jan 1, 2027: developer disclosures, deployer notice, adverse-outcome explanations, human review and three-year records. State enforcement is currently on hold under a federal court order.
IllinoisHuman Rights Act amendment (HB 3773, in force Jan 1, 2026): no AI with discriminatory effect, no zip codes as a proxy for protected classes, notice of AI use. Separate consent rules for AI video interviews and biometrics.
New JerseyDisparate-impact rules in force since Dec 15, 2025 cover automated tools, including vendor-supplied ones.
ConnecticutUsing an automated tool is no defense to a discrimination claim; notice and explanation duties for AI hiring tools start Oct 1, 2027.
TexasTRAIGA (in force Jan 1, 2026) bans AI used with intent to discriminate; disparate impact alone is not enough. Attorney General enforcement only.
European Union
The EU AI Act classifies AI used for recruitment, selection, promotion, termination, task allocation and performance monitoring as high-risk. After the 2026 Digital Omnibus, those obligations apply from December 2, 2027. Providers (developers) and deployers (employers) carry separate duties, and the Act reaches any tool placed on the EU market, wherever the vendor is based. See our EU AI Act explainer.
Rest of the world
JurisdictionWhat applies
United KingdomNo AI statute. The Equality Act 2010 sets protected characteristics; UK GDPR Articles 22A–22D (in force Feb 5, 2026) allow significant automated decisions only with safeguards: tell the candidate, allow challenge and provide meaningful human review. The ICO expects bias monitoring.
CanadaNo federal AI law. Ontario requires publicly advertised job postings to disclose AI used to screen, assess or select (employers with 25+ staff, since Jan 1, 2026). Quebec’s Law 25 requires notice, explanation and human review of decisions based exclusively on automated processing.
IndiaNo AI-hiring law. DPDP Act notice and consent duties apply from May 13, 2027, including a lawful basis before reusing candidate data for testing.
JapanSoft-law AI governance, but strict rules on what candidate data may be collected at all. Race, family origin, religion and similar attributes ordinarily cannot be collected, so testing often relies on synthetic data.
SingaporeNo AI-hiring law yet. The Workplace Fairness Act is targeted for end-2027 for employers with 25+ staff. The data regulator’s AI guidelines expressly allow personal data to be used for bias assessment.
AustraliaNo AI Act. From Dec 10, 2026, privacy policies must disclose decisions made solely or substantially by automated means; the regulator signals that a human final step is not a safe harbour.
UAENo federal AI law. DIFC Regulation 10 governs autonomous systems for DIFC-registered entities; mainland labour law prohibits discrimination in hiring.
Dates to plan around
- Dec 10, 2026: Australia’s automated-decision disclosure duty.
- Jan 1, 2027: California CCPA automated-decision duties; Colorado SB 26-189 (enforcement on hold).
- May 13, 2027: India DPDP core obligations.
- Oct 1, 2027: Connecticut notice duties for AI tools that interact with applicants.
- End 2027 (target): Singapore Workplace Fairness Act.
- Dec 2, 2027: EU AI Act high-risk obligations for hiring AI.
This is a planning map, not legal advice. Status as of November 2026; confirm each jurisdiction with local counsel.
Map your hiring AI against the laws that apply to you
Whether you build HR AI or deploy it, Fairly AI tests it, monitors the rules where you operate and keeps the evidence.
Schedule a discovery call