> This is the markdown version of https://www.classet.ai/blog/california-admt-hiring-rules > Learn more at https://www.classet.ai ![Low-poly illustration of four faceted gold paths converging on one point beneath a geometric balance scale](/_next/image?url=%2Fimages%2Fblog%2Fcalifornia-admt-hiring-rules.png&w=3840&q=75) # California ADMT Rules for Hiring: What Lands on January 1, 2027 California's ADMT regulations and Colorado's rewritten AI Act both reach hiring on January 1, 2027. What each one asks of employers, and why the cheapest compliance move is a human who actually decides. [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=96&q=75) Paul JonesHead of Growth at Classet ](/blog/authors/paul-jones) October 7, 2026 AI Recruiting, Guides & Insights On January 1, 2027, two state rulebooks reach your hiring funnel on the same day. California's privacy regulator starts enforcing its rules on automated decisionmaking technology (ADMT), and Colorado's rewritten AI Act takes effect. If your recruiters use an AI match score, an auto-reject rule, or a screening tool inside the ATS, both can apply to candidates you already have in the pipeline. Most teams I talk to are treating this as a paperwork project, but I think the bigger lever is who makes the decision and how. Quick answer California's CCPA regulations on automated decisionmaking technology took effect January 1, 2026, and employers using ADMT for significant decisions like hiring must comply by January 1, 2027. That means a pre-use notice, an opt-out or a human appeal, and a risk assessment. ADMT covers technology that replaces or substantially replaces human decisionmaking. When a person with real authority reviews the output and makes the call, much of that exposure shrinks. Talk to counsel about your setup. ## What are California's ADMT regulations? They're rules issued by the California Privacy Protection Agency (CPPA) under the California Consumer Privacy Act. On September 23, 2025, the [CPPA announced](https://cppa.ca.gov/announcements/2025/20250923.html) that the state's Office of Administrative Law had approved the package, which covers ADMT, risk assessments, and cybersecurity audits. The dates that matter for hiring: - **January 1, 2026:** the regulations took effect, and businesses subject to the risk assessment rules had to start complying. - **January 1, 2027:** businesses that use ADMT to make significant decisions must comply with the ADMT requirements. - **April 1, 2028:** businesses submit to the CPPA an attestation that required risk assessments were completed, plus a summary of them. Because the CCPA covers job applicants and employees, the rules reach HR. Per [Littler's summary](https://www.littler.com/news-analysis/asap/californias-automated-decisionmaking-technology-regulations-seven-steps), the "significant decisions" in employment are hiring, allocation of work, compensation, promotion, demotion, suspension, and termination. ## When does a hiring tool count as ADMT? The definition turns on whether the technology replaces human decisionmaking or substantially replaces it. A tool that ranks, scores, or filters candidates can qualify if its output decides who advances and nobody meaningfully checks it. Littler describes what "meaningful human review" requires. The reviewer has to: 1. Know how to interpret and use the tool's output. 2. Actually review and analyze that output, along with any other relevant information. 3. Have the authority to make or change the decision. All three have to hold in practice. A recruiter who clicks "approve" on a pre-sorted reject list, with no time and no authority to overrule it, doesn't count. ## What do employers owe if they use ADMT in hiring? If a tool does count, you owe candidates a few things. **A pre-use notice.** Before the tool processes a candidate's information, you explain what it does, what categories of personal information it uses, what kind of output it produces, how that output feeds the decision, and how to opt out, with at least two ways to submit a request. **An opt-out, or a human appeal instead.** Candidates can ask to opt out of ADMT. There are narrow exceptions. The one most employers will look at is the human appeal exception: you can skip the opt-out if you offer an appeal to a human reviewer who can interpret the output, reviews it with other relevant information, and has authority to overturn the decision. **A risk assessment.** Before you use ADMT for a significant decision, you document the purpose, the data, the risks to candidates, and the safeguards. Littler notes HR staff need to take part, the assessment gets reviewed at least every three years, and it gets updated within 45 days of a material change. ## Why are auto-reject pipelines the real exposure? Look at where AI already sits in most enterprise ATS setups: a match score at the top of the requisition, a knockout rule that moves people to "Not Selected" overnight, a ranking that decides which 40 of 600 applicants a recruiter ever opens. In those setups software makes the call, and that's where the recent lawsuits point. - **Mobley v. Workday.** On June 22, 2026, Judge Rita Lin largely denied Workday's motion to dismiss the third amended complaint, [according to Legal500](https://www.legal500.com/intelligence/united-states/employment-and-hr/recent-developments-in-mobley-v-workday-california-court-allows-key-ai-hiring-bias-claims-to-move-forward). The California anti-discrimination claims and the ADA claim go forward into discovery. We cover the full case history in [our Workday lawsuit breakdown](/blog/workday-ai-hiring-lawsuit). - **Kistler v. Eightfold AI.** Filed in January 2026, this putative class action argues that Eightfold's Match Scores are consumer reports under the Fair Credit Reporting Act, which would bring notice and dispute rights with them ([ZwillGen summary](https://www.zwillgen.com/litigation/plaintiffs-allege-ai-hiring-tools-violate-fcra/)). Eightfold denies it's a consumer reporting agency. As of this writing, no court has ruled on the merits and no class has been certified. California also has a second rule in play. The Civil Rights Council's amendments to the Fair Employment and Housing Act regulations, [effective October 1, 2025](https://www.paulhastings.com/insights/client-alerts/new-california-regulations-on-employers-use-of-ai-to-make-decisions-go-into-effect-oct-1-2025), make clear that automated-decision systems can create discrimination liability, extend that liability to an employer's agents, and require four years of record retention for data used in or resulting from those systems. They also treat evidence of anti-bias testing as relevant to a claim or a defense. ## What does Colorado add on the same date? Governor Polis signed SB 26-189 on May 14, 2026. It rewrote Colorado's AI Act, which now takes effect January 1, 2027. The rewrite dropped impact assessments and the duty of care. What's left for employers, [per Epstein Becker Green](https://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers): clear notice before using ADMT for a consequential decision, a plain-language explanation within 30 days of an adverse outcome, a chance for meaningful human review and reconsideration when commercially reasonable, and three years of records. The Attorney General enforces it. Our [Colorado AI Act guide](/blog/colorado-ai-act-hiring) goes through each piece. ## What should you have ready by January 1, 2027? This is a planning list, not legal advice. Your counsel decides what applies to you. Obligation California ADMT (CCPA) Colorado (SB 26-189) What to have ready Notice before use Pre-use notice: purpose, data, output, how it's used, opt-out Clear and conspicuous notice before a consequential decision Notice language on job postings and careers pages, by tool Candidate choice Opt-out, or human appeal exception No opt-out; human review and reconsideration when commercially reasonable An appeal route staffed by someone who can overturn the outcome After a rejection Human appeal, if you rely on that exception Plain-language explanation within 30 days A template you can fill in for a real rejected candidate Assessment Risk assessment before use; attestation by April 1, 2028 Removed by SB 26-189 Inventory of every tool that scores, ranks, or filters Records Four years under the FEHA regulations Three years ADMT inputs and outputs on the candidate record in your ATS ## Where does Classet fit? [Classet ATS Sync](/product/joy-sync-ats-plugin) puts AI phone interviews inside the ATS you already run, and it's built around a person making the decision. Joy, Classet's AI recruiter, calls candidates the moment they apply, asks your questions, and sends your team the summary, recording, and transcript. Scoring is transparent per criterion. Knockouts are your must-haves, marked pass or fail. Fast-track marks each criterion met or not met with the candidate's own quote, then gives a fit of Strong, Good, Borderline, or Weak. Joy reads only the call transcript, never a résumé or the sound of someone's voice. Joy never rejects anyone. Nothing is rejected by software, and your team decides who moves forward. The summary, transcript, and recording write back to the candidate record in the ATS, so the inputs and outputs of each screen sit where your records already live. If you configure knockout results to move candidates to a stage, that's a rule your team sets, and worth reviewing with counsel against the definitions above. Warden AI runs a [monthly third-party bias audit](/blog/ai-recruiting-bias-audits-compliant-platform) on our scoring. None of this makes an employer compliant on its own; your notices, appeal process, and risk assessment are still yours to write. The design keeps a reviewer with authority in the loop with the evidence in front of them, which both states ask for. ## Common questions ### Does human review take a tool out of the ADMT rules? It can, if the review is meaningful. The reviewer needs to know how to interpret the output, actually review it alongside other relevant information, and have authority to make or change the decision. A rubber-stamp approval of a software-sorted list won't meet that bar. Confirm how it applies to your workflow with counsel. ### Do candidates get to opt out of AI screening in California? Under the CCPA regulations, candidates can request to opt out of ADMT used for significant decisions, unless an exception applies. The human appeal exception lets an employer offer an appeal to a qualified human reviewer with authority to overturn the decision instead. ### Is Colorado's AI Act still in effect for employers? Yes, in narrower form. SB 26-189 rewrote it and moved the effective date to January 1, 2027. Employers keep notice, a post-decision explanation within 30 days, human review when commercially reasonable, and three years of records. See our [Colorado AI Act guide](/blog/colorado-ai-act-hiring). ### Does using Classet make us compliant with ADMT rules? No vendor can make you compliant. Classet is designed so a recruiter makes every hiring decision with the summary, transcript, recording, and per-criterion results in front of them, and those records write back to your ATS. Your notices, appeal process, and risk assessment still need to come from your team and counsel. ## Key points - California's ADMT compliance date for significant decisions, including hiring, is January 1, 2027, the same day Colorado's rewritten AI Act takes effect. - ADMT covers technology that replaces or substantially replaces human decisionmaking, so how your team makes the decision matters as much as which tool you buy. - Meaningful human review needs a reviewer who understands the output, actually reviews it with other information, and can change the decision. - Auto-reject rules and match scores inside the ATS are where the legal pressure sits, from Mobley v. Workday to the FCRA claims in Kistler v. Eightfold. If you want to see how a recruiter-decides AI phone interview runs inside your ATS, [request a demo](/demo?ref=blog-california-admt-hiring-rules-bottom). [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=128&q=75) Paul Jones Head of Growth at Classet Paul comes from an operator background running an Alpine-owned company, and brings firsthand experience with the hiring challenges Classet was built to solve. He's driven by a belief that the right technology can make meaningful work more accessible. ](/blog/authors/paul-jones) Follow Classet in Google Add classet.ai as a preferred source and our hiring research is more likely to show up in your Top Stories and AI Overviews. [Add as preferred source →](https://www.google.com/preferences/source?q=classet.ai) ## Explore More ### Use Cases - [RPO / BPO Recruiting](/use-cases/call-centers-bpo) - [Data Centers Recruiting](/use-cases/data-centers) - [Healthcare Recruiting](/use-cases/healthcare) ### Integrations - [Greenhouse](/integrations/greenhouse) - [Bullhorn](/integrations/bullhorn) - [Lever](/integrations/lever)