AI Regulation Tracker / For HR and Executives
California's AI Layoff Order: New WARN Act Tripwires
California just told its agencies to rewrite the rules for AI-driven layoffs, and HR leaders have a short window to prepare before the obligations arrive.
On May 21, 2026, California Governor Gavin Newsom signed Executive Order N-6-26, directing state agencies to recommend updates to the California WARN Act, severance standards, collective bargaining, and worker training tied to AI-driven job displacement. The Order imposes no new binding duties on private employers yet; it directs agencies to deliver recommendations on a 90 to 180 day timeline through October 2026.
What California actually did, and what it did not do
On May 21, 2026, Governor Gavin Newsom signed Executive Order N-6-26, which the Governor's office called a first-of-its-kind move to prepare workers and businesses for AI disruption. Read the fine print and you find something more modest than the headlines suggest, and more important. The Order does not create a single new obligation for private employers today. It tells a network of state agencies to go study the problem and come back with concrete recommendations on severance, the California WARN Act, collective bargaining, and worker training within a matter of months.
That gap between "signal" and "law" is the whole story for HR. You are not out of compliance with anything right now. You are being told, clearly, where California intends to point its regulatory machinery next. The employers who use this window well will look a lot smarter in a year than the ones who wait for the rules to land.
The four review streams that should be on your radar
The law firm K&L Gates, which published a detailed employer alert on the Order, laid out the deadlines that matter. Four of them touch HR directly.
- Cal-WARN Act review (within 180 days). The Labor and Workforce Development Agency, known as LWDA, must give the governor recommendations to update the WARN Act so it provides better early warning on emerging industry trends. Translation: expect a push to make AI-driven reductions trigger notice obligations that the current statute may not cleanly capture.
- Severance and safety-net review (within 180 days). LWDA is also studying severance and other compensation for displaced workers, including stock or equity, with a comparative look at what other countries do. Whatever comes out of this may inform future legislation on what a fair AI-displacement package looks like.
- Collective bargaining review (by October 15, 2026). The state wants to learn how unionized workplaces are handling AI at the bargaining table. If you have a represented workforce, assume AI adoption becomes a more active subject of negotiation.
- Workforce training and an AI playbook (by October 15, 2026). LWDA and the Employment Development Department are building dislocated-worker strategies and an AI playbook for exposed occupations. New training frameworks can eventually attach to employer obligations or incentives.
There is also an early-warning piece worth watching. Within 90 days, the Employment Development Department is supposed to launch a dashboard showing AI's impact on employment by sector, using unemployment insurance data and, potentially, data published by leading AI labs. That dashboard is how the state plans to spot the disruptions it will later regulate.
Why the WARN Act is the sharpest edge here
California already has a WARN Act that is tougher than the federal version. It generally requires 60 days of advance notice for covered plant closings, mass layoffs, and relocations, and it reaches smaller reductions than the federal law does. The problem the state is trying to solve is that AI-driven change does not always look like a classic "mass layoff." It can look like slow attrition, role consolidation, or a reorganization that quietly removes the need for a job function without a single dramatic announcement.
If LWDA recommends, and the legislature later adopts, a WARN Act that captures AI-related workforce changes more aggressively, the practical effect is this: reductions you would not have flagged for notice today could carry a notice duty tomorrow. Missing a WARN obligation is expensive. Back pay and benefits for the notice period, plus civil penalties, add up fast across a large reduction. That is why the smart move is to build the muscle for notice analysis now, while the stakes are still hypothetical.
How to prepare now: a documentation and decision framework
None of this requires you to change your headcount plans. It requires you to change how you record and defend them. Here is a concrete checklist you can put in front of your HR and legal teams this quarter.
1. Create an AI-restructuring log. For every role change where AI played a part, capture four things: what the AI tool actually does, what human made the final call, the business reason for the reduction, and the date. Regulators and plaintiffs both reward a clean paper trail and punish a vague one. 2. Keep a human in the decision. California's Civil Rights Council already put its FEHA automated-decision-system regulations into effect on October 1, 2025, requiring accountability, bias testing, and recordkeeping when a covered automated-decision system factors into employment decisions. Do not let an AI system be the sole reason a person loses a job. Document the human judgment on top of it. 3. Pressure-test your WARN triggers. Run your planned reductions against current Cal-WARN thresholds now, and re-run them assuming a broader definition that counts AI-driven consolidations. Know where you would be exposed if the rules tighten. 4. Pre-draft a displacement package standard. Decide, before you are under deadline pressure, what your severance, notice, and retraining support look like for AI-related reductions. A policy you designed calmly beats one you improvise during a layoff. 5. Map your union exposure. If any part of your workforce is represented, get ahead of the bargaining question. Identify where AI adoption might be a mandatory subject and prepare your position. 6. Assign an owner. One named person should track the agency reports as they publish through late 2026. This Order is a calendar, not a one-time event.
Managers who can explain, in plain language, why a restructuring happened and what role AI played will be worth their weight this year. That is a training gap for most teams, and it is one worth closing. If you are building that capability, our [AI for Managers course](/ai-for-managers) is designed for exactly this kind of decision-making, and you can [find the right starting point in about two minutes](/quiz).
The honest read on timing
It is tempting to file this under "future problem." Resist that. The Order sets 90-day and 180-day clocks, and several deliverables are due by October 15, 2026. California has a habit of moving from study to statute quickly once the groundwork is laid. The agencies are not asking whether to regulate AI-driven displacement. They are working out how. Employers who treat the next few months as preparation time, rather than waiting time, will not be scrambling when the recommendations turn into bills.
The bigger pattern
California is not acting alone, and that is the part multi-state employers cannot ignore. Connecticut passed a broad AI law, SB 5, that regulates high-risk AI systems, including automated employment decision technology, with obligations phasing in between October 2026 and January 2028. Colorado overhauled its AI Act: after pushing the original SB24-205 effective date to June 30, 2026, the state replaced it with SB 26-189, which takes effect January 1, 2027. Illinois amended its Human Rights Act through HB3773 to bar discriminatory AI use and require employee notice, effective January 1, 2026, and New Jersey's Division on Civil Rights has issued guidance applying its Law Against Discrimination to automated employment decisions. The result is a growing patchwork where the same reduction can carry different obligations depending on where your workers sit. A single national policy for AI-driven workforce change is starting to look less like good practice and more like a survival requirement.
Frequently Asked Questions
Does Executive Order N-6-26 create new WARN Act duties I have to follow right now?
No. The Order directs the Labor and Workforce Development Agency to recommend WARN Act updates within 180 days. It does not change the statute itself. Your current Cal-WARN obligations are unchanged until the legislature or a regulator acts on those recommendations.
We only have a handful of California employees. Does this reach us?
Possibly, over time, but check the threshold first. Cal-WARN applies to covered establishments with 75 or more employees, counting current and recent staff, so the smallest employers sit outside it today. That said, California's WARN Act already covers smaller reductions than federal law, and the Order aims to make it more responsive to AI-driven change. Even small employers should track the recommendations, because thresholds and triggers are exactly what the state is reviewing.
Can we let an AI tool make the layoff decision if a human signs off at the end?
Be careful here. California's FEHA automated-decision-system regulations, effective October 1, 2025, already require accountability and recordkeeping when a covered automated-decision system factors into employment decisions. Keep a genuine human judgment in the loop and document the reasoning, not just the AI output.
What is the single most useful thing to do before the recommendations publish?
Start a written log for every AI-influenced role change now. Capture what the AI did, who made the final call, the business reason, and the date. If notice or disclosure rules expand, that record is your best defense and your fastest path to compliance.
When should we expect actual rules, not just reports?
The Order sets deadlines running from 90 days to October 15, 2026, for the underlying reports. Legislation or formal rulemaking would come after that. There is no fixed date for binding employer duties yet, which is precisely why the current window is useful.
Does this affect union and non-union employers differently?
Yes. One review stream specifically studies how collective bargaining is handling AI adoption. Employers with represented workforces should expect AI to become a more active bargaining subject, while non-union employers should focus on documentation and WARN readiness.
Sponsored Training
Browse the full AI Regulation News tracker
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.