AI for HR in a Small Business: A Traffic-Light Guide, Plus California's New Rules

"Nobody wants to be fired by a machine," said state Sen. McNerney, author of California's new No Robo Bosses Act. From July 1, 2027, a California employer that lets software alone decide to discipline or fire someone risks a $500 civil penalty per violation. Most HR uses of AI sit nowhere near that line. Here is which tasks are safe to hand over, which need a human check, and which are now off-limits. This is general information, not legal advice.

The Traffic-Light Table

The line between yellow and red is a score. A plain summary of a resume you read yourself is a writing aid. A tool that rates, ranks or flags people is making part of the decision, and that is where the laws start to bite. If you paste staff details into a chatbot, read what happens to that data first.

California's Two New Rules, Plainly

Gov. Newsom signed both bills on September 30, 2026. None of the law-firm analyses we reviewed describe a small-employer exemption for either one.

SB 947, the No Robo Bosses Act (from July 1, 2027). It covers an "automated decision system": in Reed Smith's summary of the law, "any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence" that issues "a score, classification, or recommendation." Spam filters and antivirus software are carved out. The rules:

• You may not rely solely on such a system to discipline or fire an existing employee. Fisher Phillips notes it does not apply to job applicants.
• If you primarily rely on it, a human must corroborate the output. If the reviewer finds it "inaccurate, incomplete, or misleading," you cannot use it.
• When you tell the employee the decision, you must give a written, plain-language notice that you relied on the system, and, on request, a "meaningful, objective description" of their data it used.
• Penalty: $500 per violation, enforced by the Labor Commissioner and public prosecutors. Morgan Lewis and Reed Smith say the law creates no individual right to sue, though Reed Smith expects plaintiffs to try anyway under California's existing employee penalty law.

AB 1883 (from January 1, 2027). Employers may not use AI workplace surveillance tools to recognize or predict an employee's emotional state, or to collect neural data, meaning measurements of the nervous system's activity. Hunton notes a limited exception for AI tools used to ensure safety. Morgan Lewis says it "applies broadly to private and public employers." Penalties run up to $500 per violation.

A Rubber Stamp Will Not Count

The human check has to be real. "The test is whether the human checked the evidence and had the power to say no," Frank Dickson of Dickson Research told CIO. A manager who forwards the software's verdict is not reviewing it. The reviewer should look at the underlying records, such as personnel files and past evaluations, and be free to overrule the tool.

What to Do Before 2027

1. List every tool that scores, ranks or flags employees, including scheduling, sales and monitoring software. AI often hides in tools bought for other reasons.
2. Check monitoring tools for mood or emotion features and switch them off before January 1.
3. Name your reviewer, and keep a note of what they checked for each discipline or firing decision.
4. Draft the notice now: a short standalone letter saying the system was used, a human reviewed it, and who to contact.
5. Ask vendors to show the data behind each score. If they cannot, do not use the tool for discipline.

Hiring in Colorado? A separate law starting January 1, 2027 covers AI in employment decisions, including for applicants; see our Colorado AI Act guide. And if your green-light tasks still eat hours each week, that is the kind of chore our workflow automation projects take off your plate.

Sources

1. KQED, "Newsom signs slate of AI workplace laws barring 'robo bosses' and surveillance" (October 1, 2026): signing, July 1, 2027 start, McNerney quote. kqed.org
2. HR Dive, "California revamps AI protections for workers in flurry of bill signings" (October 1, 2026): SB 947 and AB 1883 summaries. hrdive.com
3. Fisher Phillips, "California Just Banned Employers From Letting AI Fire Workers on Its Own: Your 5-Step Compliance Plan" (October 1, 2026): existing employees only, carve-outs, notice timing. fisherphillips.com
4. Covington, "New California AI Employment Laws: Key Requirements for Employers" (October 2, 2026): operative dates and penalties. insideprivacy.com
5. Hunton, "California Enacts New Restrictions on AI-Based Employee Monitoring" (October 6, 2026): AB 1883 scope, safety exception, enforcement. hunton.com
6. Morgan Lewis, "California Enacts New Laws Governing Employer Use of AI" (October 9, 2026): requirements, enforcement, coverage. morganlewis.com
7. Reed Smith, "What's New in California: A Closer Look at Recently Enacted Senate Bill 947" (October 12, 2026): system definition, penalty, no private right of action. reedsmith.com
8. CIO, "A human rubber stamp on AI-based HR decisions won't satisfy California's No Robo Bosses law" (October 1, 2026): Dickson quote, review practices. cio.com
9. Morrison Foerster, "Colorado Hits Reset on AI Regulation With a New AI Act" (May 15, 2026): January 1, 2027 start; employment and applicants covered. mofo.com