AI hiring laws: what employers must do before the 2027 deadlines

In three weeks, California signed a ban on AI-only firings, Connecticut's new AI law switched on, and Colorado is days away from a public hearing on the rules for its own statute. If your company uses software to screen résumés, rank applicants or flag performance problems, AI hiring laws are no longer a someday issue. The biggest deadlines land on January 1, 2027, about ten weeks from now.

None of these laws bans AI from the hiring process. What they do is set conditions: tell people when a tool is involved, test it for bias, keep records and make sure a human being can stand behind the decision.

What changed this fall

California. On September 30, Governor Newsom signed SB 947, the "No Robo Bosses Act," a year after vetoing an earlier version. HR Dive reports that it bars employers from relying solely on an automated decision system to discipline or fire someone. When a tool drives those decisions, a person must corroborate the result with other information, such as evaluations and personnel files. Workers get written notice and can ask what data the system used. It takes effect July 1, 2027. The same batch of signings included SB 951, which requires mass-layoff notices when cuts are caused wholly or substantially by AI.

Connecticut. Public Act 26-15 took effect October 1. According to Mintz, employers can no longer point to an automated employment decision tool as a defense to a discrimination complaint, though evidence of anti-bias testing may still be considered. Layoff notices filed with the state labor department must now say whether the cuts relate to AI. The heavier disclosure duties start October 1, 2027, and the attorney general enforces them.

Colorado. The state repealed its original AI Act and replaced it with SB 26-189, which takes effect January 1, 2027. The attorney general filed proposed rules in August, and Law Week Colorado reports that the public hearing and written-comment deadline are both October 26. The draft rules tackle two questions every employer will care about: when an AI output "materially influences" a decision, and what you owe when you rely on a vendor's tool.

AI hiring laws already in force

Several rules apply today, and the October 2026 Gibson Dunn newsletter on AI in employment is a useful map of them:

• New York City (Local Law 144): Since 2023, employers using an automated employment decision tool must have a bias audit done within the year before use, post a summary of results and give candidates notice at least 10 business days ahead. A December 2025 state comptroller audit found the city had flagged one noncompliant company out of 32 surveyed, while the comptroller's own reviewers spotted 17 potential problems.
• Illinois (HB 3773): Since January 1, 2026, AI that has a discriminatory effect in employment decisions is prohibited, notice of AI use is required, and zip codes cannot serve as a stand-in for a protected class. State rules were issued and then withdrawn this year, with no new timeline.
• California civil rights rules: Since October 1, 2025, regulations under the state's fair employment law cover automated decision systems, extend to agents acting for an employer and treat anti-bias testing as relevant evidence. Records must be kept four years.
• New Jersey and Texas: New Jersey's disparate-impact regulations explicitly cover automated tools. Texas's law targets intentional discrimination only and adds no notice or audit duties.

Software can sort the pile, but a person still owns the decision.

January 1, 2027: the big date for AI in hiring

Two sets of obligations arrive on New Year's Day.

California privacy rules. Under regulations from the California Privacy Protection Agency, businesses covered by the state's privacy law that use automated decision-making technology for significant employment decisions must comply by January 1, 2027. HR Dive summarizes the duties as pre-use notices, a chance to opt out and completed risk assessments. Gibson Dunn notes penalties of up to $2,663 per violation and $7,988 per intentional violation.

Colorado's rewritten law. Employers using covered tools will owe applicants and employees notice before use, an explanation within 30 days after an adverse outcome, and the ability to correct data and request human review. The law also bars contract clauses that indemnify a developer or employer for discriminatory use of a tool. A federal lawsuit over Colorado's law is pending, but employers should not plan around a court rescuing them.

Is there a federal rule?

Not yet. There is no federal statute on AI in employment. A December 2025 executive order directed agencies to challenge and preempt state AI laws, and the Justice Department set up an AI litigation task force in January 2026. Until Congress acts or a court strikes something down, the state and city AI hiring laws above apply as written, and federal anti-discrimination law still covers decisions made with any tool.

What employers should do this quarter

You do not need a compliance department to get ahead of this. Start with these steps, and talk to an employment lawyer about which rules reach your workforce.

1. List every tool. Include résumé screeners, applicant-tracking rankings, chatbots, video-interview scoring, scheduling software and performance flags. Many teams find automated decision tools built into software they already use.
2. Map your people. Most of these rules turn on where applicants and employees are located, so note which states and cities you hire in.
3. Press your vendors. Ask for bias-audit results, documentation of how the tool works and what data it uses. Review contracts for indemnity terms that Colorado's law prohibits.
4. Draft notices now. Plain-language notices for California, Colorado and New York City take time to write and approve. Do not wait for December.
5. Make human review real. The person signing off should be able to override the tool and should write down why. A reviewer who simply clicks "approve" will not satisfy California's new standard.
6. Keep records. Save audit results, notices, decisions and reasons. California already requires four years.

Job Elf tip: For each tool, write one sentence that names the decision it influences and the person who signs off. If you can't fill in the name, that step needs a human owner before January 1.

Hiring that starts with a person reading real experience is easier to explain and defend. On Job Elf, employers can search candidate profiles directly and watch short video introductions, and our employer resources cover fair, practical hiring.

The bottom line

AI hiring laws are a patchwork, but they point the same way: disclose, test, document and keep a human accountable. Use the next ten weeks to inventory your tools, lean on your vendors and get notices ready, so January 1 is a normal workday rather than a scramble.