You're using ChatGPT to screen resumes. Your ATS ranks candidates automatically. You're considering an AI interview tool that "predicts job fit." A friend tells you: "Be careful, there are AI hiring laws now." You think: "Is that real or just hype?"

Most founders think: "AI hiring laws are coming someday. I'll worry about it when they're actually enforced." What they don't realize: AI hiring laws are already in effect in multiple states right now. Companies are being audited, fined, and sued. If you're using AI to screen, rank, or assess candidates in California, Colorado, Illinois, New York, or Texas, you're subject to specific legal requirements, whether you know it or not.

The expensive truth: Illinois now allows candidates to sue you directly if your AI hiring tool discriminates, even unintentionally. California holds you liable for your vendor's AI discrimination. Colorado requires impact assessments and transparency notices starting June 30, 2026. Penalties range from $20,000 per violation to 7% of global revenue under EU rules (if you hire internationally). "I didn't know" is not a defense.

Here's what AI hiring laws exist right now in March 2026, what they require, and how to stay compliant.

Federal status: No comprehensive federal AI employment law exists. Trump's December 2025 Executive Order seeks to preempt state laws, but the Senate voted 99-1 to reject an AI moratorium. State laws remain fully enforceable.

What this means: You must comply with state-by-state requirements. Multi-state hiring means multi-state compliance.

These laws regulate "Automated Employment Decision Tools" (AEDTs) or "Automated Decision Systems" (ADS):

What's NOT covered: Manual review of resumes, human-only interviews, basic email/calendar scheduling tools

Critical point: Even if a human makes the final decision, if AI influenced that decision (screened, scored, ranked), you're subject to these laws.

Who it applies to: Employers and employment agencies using AEDTs in NYC

Requirements:

☐ Annual bias audit: Independent third-party must test AEDT for discriminatory impact on race/ethnicity and sex. Audit must be conducted within one year before use.

☐ Publish audit results: Make bias audit results publicly available on website (selection rates by race/ethnicity and sex). Even if you use a third-party AEDT, you need to make sure they are audited and post their audit results on your website.

☐ Notice to candidates: Inform candidates at least 10 days before use that AEDT will be used, what job qualifications it assesses, and what data sources it uses

☐ Alternative process: Provide accommodation for candidates who request alternative selection process

Penalties: NYC is actively auditing compliance Consultils. Violations can result in civil penalties.

Example - 15-person startup, New York: Uses AI resume screening tool. Must: (1) Get annual bias audit from independent auditor, (2) Post audit results on website showing selection rates by race/sex, (3) Add notice to job postings: "This employer uses an automated employment decision tool (AEDT) to screen candidates. You may request an alternative selection process or accommodation."

Who it applies to: All California employers using Automated Decision Systems (ADS)

Requirements:

☐ No discriminatory ADS: Cannot use ADS that discriminates based on protected characteristics (race, sex, age, disability, etc.) Consultils

☐ 4-year data retention: Retain all ADS-related data for at least 4 years, including input data, outputs (scores/rankings), criteria used, and bias testing results Consultils

☐ Vendor accountability: You are responsible for discriminatory outcomes caused by third-party vendors using ADS Consultils. "Our vendor did it" is not a defense.

☐ Proactive testing: While not explicitly mandated, regulators will assess whether testing occurred and the testing's quality, scope, recency, results, and the employer's response to identified risks Manatt, Phelps & Phillips, LLP. Unsupported vendor assurances carry little weight.

☐ Transparency notices: Inform candidates when AI influences hiring, promotion, or employment decisions

Penalties: Enforcement through California Civil Rights Department. Violations treated as employment discrimination under FEHA.

Example - 25-person startup, California: Uses AI-powered interview platform from vendor. Startup is liable if platform discriminates, even if vendor built it. Must: (1) Retain 4 years of data on who was scored how and why, (2) Conduct or review bias testing, (3) Notify candidates AI is used.

Who it applies to: All Illinois employers (even 1-employee companies)

Requirements:

☐ No discrimination (including unintentional): Using AI that results in discrimination violates Illinois Human Rights Act, intent is not required The HR DigestManatt, Phelps & Phillips, LLP

☐ Transparency: Inform employees and candidates when AI is used to make employment decisions (hiring, promotion, discipline, firing) The HR Digest

☐ Human review: Final decisions cannot be fully automated

Critical difference: Illinois is the ONLY U.S. state with a private right of action: candidates who believe they were discriminated against through AI hiring tools can sue the employer directly in court Hirevire, without filing a government complaint first.

Penalties: Civil rights lawsuits, damages, attorney's fees. Litigation risk is highest in Illinois.

Example - 12-person startup, Texas, hiring remote workers in Illinois: Uses AI to screen candidates. Illinois candidate applies. Even though company is in Texas, they're subject to Illinois law for that candidate. If AI discriminates (even unintentionally), candidate can sue directly.

Who it applies to: "Deployers" of high-risk AI systems (employment AI is classified as high-risk)

Requirements (starting June 30, 2026):

☐ Risk management policy: Create documented policy for identifying and mitigating algorithmic discrimination Consultils

☐ Impact assessments: Evaluate high-risk AI systems to identify and mitigate potential harm K&L Gates

☐ Transparency notices: Inform candidates and employees when AI influences employment decisions (hiring, firing, promotion) K&L Gates

☐ Appeal rights: Provide mechanism for candidates/employees to appeal AI-influenced decisions

☐ Public statement: Make publicly available statement about types of AI systems used and how they manage discrimination risk

☐ Human review: Ensure final employment decisions include meaningful human review, not fully automated

Penalties: Violations deemed "unfair trade practice," civil penalties up to $20,000 per violation Consultils. Example: Failing to notify 10 rejected applicants = $200,000 in potential fines. No private right of action; Attorney General enforces.

Example - 8-person startup, Colorado: Plans to use AI interview tool. Before June 30, 2026 must: (1) Document risk management policy, (2) Conduct impact assessment, (3) Add notice to job postings, (4) Create appeal process, (5) Post public statement on website.

Who it applies to: Texas employers using AI in employment decisions

Requirements:

☐ Governance framework: Establish oversight for AI deployment

☐ Avoid discrimination: Cannot use AI that discriminates based on protected characteristics

☐ State council oversight: Subject to Texas AI advisory council guidance

Business-friendly provisions:

Example - 20-person startup, Texas: Uses AI for candidate screening. Must establish governance, avoid discrimination, but has lower litigation risk than Illinois (no private right of action, higher bar for discrimination claims).

Current status: Florida has no state-level AI employment laws in effect.

What still applies: Federal employment discrimination laws (Title VII, ADEA, ADA) apply if AI creates discriminatory outcomes. EEOC can enforce.

Future: Governor proposed AI regulations, but none specific to employment have passed yet.

Critical rule: Employers recruiting remote workers may be subject to these laws even if they are not physically located in the regulating jurisdiction Lexology.

Scenario - Startup in Florida, hiring nationally:

You need compliance for the candidate's location, not your location.

What it means: Regularly test your AI tools for discriminatory impact on protected classes (race, sex, age, disability, etc.)

How to comply:

What it means: Inform candidates when AI is used in hiring decisions

How to comply:

What it means: Final employment decisions cannot be fully automated and must include meaningful human review Lexology

How to comply:

What it means: You are responsible for discriminatory outcomes from third-party AI tools Consultils

How to comply:

What it means: Keep records of AI decisions, inputs, outputs, and testing

How to comply:

Ask:

Document: List every AI tool in your hiring process

Based on where you hire:

Request from every AI vendor:

If vendor can't provide: Consider switching vendors or conducting your own testing.

Update job postings: "[Company] uses AI-powered tools to screen candidates for this position. The AI assesses [specific qualifications]. Data sources include [resume, application responses, assessments]. Candidates may request an alternative selection process by contacting [email]."

Ensure:

Maintain:

What to Know About Hiring with AI

AI hiring laws are not coming; they're here.

As of March 2026:

Federal preemption is uncertain. Trump's executive order seeks to override state laws, but the Senate rejected an AI moratorium 99-1. Compliance with state and local AI laws remains mandatory, and employers should not assume that federal action will simplify obligations in the near term Lexology.

Three actions this week:

The risk is real. Compliance is not optional.

Illinois candidates can sue you directly. California holds you liable for your vendor's mistakes. Colorado can fine you $20K per violation starting June 30.

AI is powerful. AI is also regulated. Use it, but comply with the law.

If you're using AI in hiring, you need to understand these requirements now. Not when you get audited. Not when you get sued. Now.

This content is provided for informational purposes only and does not constitute legal advice; for guidance on your specific situation, please consult with an employment attorney licensed in your state.