Skip to content
Artificial Intelligence

California's New AI Layoff Law Will Make You Name the Robot.

California's new AI layoff law doesn't lower WARN's thresholds. It forces employers to name the technology when AI drives the cuts.

Sixty days. Seventy-five employees. Fifty layoffs in a month. If those numbers sound like old news, that is because they are the same Cal-WARN thresholds California has used for years, and SB 951, the California Worker Technological Displacement Act, does not touch them. What it does change is smaller and, I would argue, more interesting: if AI is why those layoffs happened, you now have to say so, in writing, and California will publish it.

SB 951 passed the Legislature on September 1 and is sitting on Governor Newsom's desk, where he has until September 30 to sign it. If he does, it takes effect January 1, 2027. I have used AI to build products and I have used it to cut headcount plans, and this bill is the first one I have seen that does not ask permission to use AI. It asks you to be honest about it after the fact. 

What does California's AI layoff law actually require?

If you already have to send a Cal-WARN notice, and the layoff was caused in whole or in substantial part by AI or another automated technology, that notice now needs three extra things. It states at the top that this is a technology displacement. It names the job functions being automated. It names the category or type of AI system or automating technology behind the cut.

The Employment Development Department then posts a summary of every notice it receives and publishes a quarterly statewide roundup. By January 1, 2028, EDD also owes the Legislature a report on how AI is actually reshaping hiring across California industries and regions.

Cal-WARN before and after SB 951

Existing Cal-WARN

Added by SB 951

Who is covered

Employers with 75+ employees in the past 12 months

Unchanged

What triggers notice

A mass layoff of 50+ employees in 30 days, a relocation, or a closure

Unchanged

Notice period

60 days

Unchanged

What the notice must say

Who, when, where

Add: this is a technology displacement, which job functions are automated, what category of AI or automating tech caused it

Who sees the notice

Affected workers, EDD, local workforce boards, city and county officials

Same recipients, plus EDD publishes a public summary and a quarterly statewide roundup

Penalties for violating notice requirements

Back pay and benefits for the notice period, civil penalties up to $500 a day, and a private right of action with attorneys' fees under existing Cal-WARN law

Same enforcement mechanism now covers the new content requirements

Does the AI layoff law drop the threshold to 25 employees?

You may have already read that it does. Early coverage of SB 951, back when it was introduced in February and amended through April, described a 25-employee or 25-percent-of-workforce trigger and a 90-day notice window. That version existed. It did not survive.

The bill that actually passed the Legislature on September 1 keeps Cal-WARN's original numbers: 75 employees to be a covered employer, 50 laid off in 30 days to trigger notice, 60 days to give it. If you read the February headlines and moved on, you have the wrong bill in your head. Go back and check what you are actually working from before you brief your leadership team on this. 

Why does naming the AI system matter more than the notice period?

Everyone who covered this bill early got excited about 90 days versus 60. I get it, that is a real operational difference for a severance budget. But the number that changes behavior is the one nobody argued about: you now have to name the category of AI system that did the displacing, in a filing the state posts online.

Right now, “we are using AI to become more efficient” is a line on an earnings call. It is directional, and nobody can hold you to it. Under SB 951, if AI substantially caused the layoff, that sentence becomes a filing with a named technology category attached to a headcount number, sitting in a public EDD database next to every other company's.

Vendors should feel that before employers do. Every AI workforce tool gets sold on a productivity story, not a replacement story, because nobody wants to be the name in that filing. I would bet on new liability language showing up in enterprise AI contracts within the year, the same way cybersecurity clauses did after the first wave of breach-disclosure laws. 

Is this actually a story about AI replacing jobs?

Not in the way most coverage will frame it. I have said this for years and I will keep saying it: AI does not replace people, it augments and automates specific tasks, and companies that tell themselves the replacement story usually find out the hard way that they oversold it.

Klarna is the clean example. In 2024 it froze hiring, leaned on an AI assistant for customer service, and let headcount drop 22 percent through attrition. By 2025 it was quietly hiring humans back after service quality slipped, in what people now call the Klarna Effect. Duolingo's CEO wrote an “AI-first” memo about phasing out contractors, then spent weeks walking it back after backlash, insisting the company was not laying off employees and was hiring at the same pace as before.

Both of those reversals were possible because they lived in a memo and a LinkedIn post. Once your reasoning lives in an EDD filing with your company's name and a technology category attached, walking it back gets a lot harder. SB 951 does not ask whether AI is replacing your people or augmenting them. It asks you to put your answer on the record. 

What five questions should HR ask before January 2027?

Forget the notice template. The form is not what trips people up. What trips people up is not being able to piece together the story the form wants.

One. Can you list every AI system deployed across the business right now, with a named owner for each?

Two. For each one, can you state the business case as it was written at the point of purchase, not as it gets described today?

Three. Do you know which job functions each tool currently performs, at the task level rather than the tool level?

Four. If headcount in a function drops next year, can you show what share of that is attributable to the tool versus demand, attrition, or restructuring?

Five. Who signs the notice? Not which department. Which named person.

In the AI-readiness work I do at livingHR, this is where almost every company hits the same wall: most can answer one and a half of these five questions. Questions one and two turn into an archaeological dig, because the purchases happened across four different budgets and nobody kept the original business case. Question three is usually where things unravel, since almost nobody has mapped tools to tasks, and you cannot defend a claim about job functions you never wrote down.

Question four is the real dividing line. You either tracked the decision as it happened or you did not, and there is no fixing that after the fact. Question five may look like paperwork, but it's a governance question in disguise. Does your People team have a real seat at the AI table, or is it left cleaning up after technology decisions get made somewhere else?

What should companies do before the law takes effect?

There is a version of this where SB 951 is just one more form in the compliance pile. That is not the version I would plan around. By making companies state the connection between a piece of software and a set of lost jobs, and then publishing it, California is forcing a choice that most organizations have spent three years avoiding: are you augmenting people or replacing them, and can you actually prove which one in writing?

Most companies have not picked a lane. They run both stories at once: replacement in the budget memo, augmentation in the all-hands meeting. This law makes you choose, on the record, with a technology category attached to a real number.

So before a notice template lands on your desk in 2027, get the answer to the five questions above while it is still hypothetical. The record you build now, not the one you write under a deadline, is the one that will hold up.

If you want help building that record before you need it, that is the kind of readiness work livingHR's AI-People Solutions team runs. 

FAQ

What is SB 951, the California Worker Technological Displacement Act?

SB 951 amends California's WARN Act to require extra disclosure when AI or other automated technology substantially causes a layoff already covered by Cal-WARN. It does not create a new, separate notice law. 

Does SB 951 lower California's WARN Act thresholds to 25 employees?

No. Earlier versions of the bill proposed a 25-employee or 25-percent-of-workforce trigger and a 90-day notice period. The version that passed the Legislature keeps Cal-WARN's existing thresholds: 75 or more employees, a mass layoff of 50 or more in 30 days, and 60 days' notice.

What does an SB 951 notice have to include?

If AI substantially caused a covered layoff, the notice must say so at the top, name the job functions being automated, and name the category or type of AI system or automating technology involved. 

When does SB 951 take effect?

The bill passed the Legislature on September 1, 2026, and is on Governor Newsom's desk with a September 30, 2026 deadline to sign or veto. If signed, it takes effect January 1, 2027.

What happens if an employer does not comply?

The same Cal-WARN enforcement mechanism applies: back pay and benefits for the notice period, civil penalties of up to $500 per day, and a private right of action that can include attorneys' fees. 

Anthony Onesto

We humanize work for everyone because we know it creates better outcomes for humanity and business.

Incoming Mail

Subscribe to our newsletter

Latest Articles

California's New AI Layoff Law Will Make You Name the Robot.

California's New AI Layoff Law Will Make You Name the Robot.

California's new AI layoff law doesn't lower WARN's thresholds. It forces employers to name the technology when AI drives the cuts.

Don’t Wait for December: Run a Q4 People Check-In Now

Don’t Wait for December: Run a Q4 People Check-In Now

Q4 is your last chance to shape 2026. Five questions HR leaders can use to run a people check-in now and finish the year strong.

AI ROI Metrics: Why 95% of AI Pilots Show No Return

AI ROI Metrics: Why 95% of AI Pilots Show No Return

MIT found 95% of AI pilots show no ROI. The real problem is not AI, it is that most companies never built a baseline to measure it.