SB 951 Turns Every California AI Layoff Into a Pre-Tagged List

On September 30, 2026, Governor Newsom signed SB 951, amending the Cal/WARN Act to force any California employer running a mass layoff caused "in whole or in substantial part" by AI to name the system, the automated job functions, and the exact occupations and work locations affected. The EDD will publish both individual notices and a quarterly statewide summary. For anyone whose job is finding people, this is not a worker-protection story. It is a government-mandated, pre-labeled target list, and it goes live January 1, 2027.

What SB 951 actually requires employers to disclose

SB 951 adds four new fields to any Cal/WARN notice tied to AI-driven layoffs, and routes a copy to the EDD for public republication. The 60-day advance notice period and existing Cal/WARN coverage thresholds survived from earlier drafts; the 90-day / 25-worker expansion did not.

Per the statute, a qualifying notice must now carry:

1. The number, classification or occupation, and work location of layoffs substantially due to AI or automation.
2. The job functions of replaced workers that will be automated.
3. The specific category or type of AI system or other automating technology that substantially caused the displacement.
4. The statement "This notice is for a technology displacement" prominently at the top of the notice.

Authored by State Senator Eloise Gómez Reyes and sponsored by the California Federation of Labor Unions, SB 951 is part of the broader AI workplace package Newsom signed the same day. The trigger standard, "in whole or in substantial part," is intentionally undefined. That ambiguity is the sourcer's friend; more on that below.

449,838

Why this is a sourcing event, not a policy event

SB 951 is the first US law that compels employers to publicly skills-tag the workers they lay off. Previous Cal/WARN notices gave you a company, a date, a headcount, and sometimes a title list. SB 951 adds verified job functions and a named technology category, which collapses two weeks of sourcing diligence into a copy-paste.

Think about what that unlocks. If Oracle files a notice saying 240 "L1 technical support" roles in Pleasanton were displaced by an "autonomous customer-service agent," you no longer have to guess which of those 240 people handled ticket triage versus escalation versus billing. The state has told you. Every laid-off worker on that notice has verified, employer-attested experience in exactly the functions named. You can route the whole cohort to a competing CX vendor's hiring pipeline the same afternoon.

That is the exact gap Refolk closes between the notice landing and your shortlist being ready: you describe the displaced cohort in plain English (role, employer, metro, timeframe) and get a ranked list across GitHub, LinkedIn, and the open web, not a Boolean you have to debug.

The three pools SB 951 will tag first

Three function clusters will dominate the first year of SB 951 filings, because they are the three where current-generation AI systems are already in production: customer support, software engineering, and content writing. Refolk's index gives us the US ceilings for each.

The 8.2x ratio is the number to internalize. Every quarterly EDD summary will skew heavily toward support, because that is where the production AI deployments are. If your desk recruits engineers only, you will see maybe one qualifying notice a month. If you recruit CX, you will see one a week by Q2 2027.

Why support leads engineering by an order of magnitude

Support leads for a boring reason: unit economics. An LLM agent that resolves a $4 ticket pays for itself immediately, while a coding agent that writes a six-figure-salaried engineer out of a job has to clear a much higher confidence bar before any CFO signs off on the RIF. SB 951 will catalog this gap in near-real-time, quarter by quarter. The quarterly statewide summary is the more valuable artifact than any individual notice, because it lets you see which AI vendors are displacing which functions fastest, and target the still-employed workers at peer companies before their RIF lands.

Field #3 is a competitive-intelligence leak

The requirement to name "the specific category or type of AI system or other automating technology" turns every SB 951 notice into a vendor-stack disclosure. This is the field most employers' legal teams have not yet reckoned with.

If a mid-market SaaS company files a notice saying its 60-person L1 support team was displaced by an "autonomous customer-service agent," rival vendors can:

• Infer the deployed vendor by cross-referencing public case studies, press releases, and the company's Zendesk or Salesforce integration history.
• Map which companies have crossed the automation threshold and which are still evaluating.
• Target the still-employed CX leads at the next ten companies on that vendor's logo page.

For sourcers working on behalf of the AI vendors themselves - Sierra, Decagon, Cursor, Cognition - SB 951 is also a lead-gen stream for implementation engineers, forward-deployed engineers, and solutions architects, because every disclosure tells you which category of system a buyer just committed to in production.

SB 951 is the first US law that forces employers to publicly skills-tag the workers they lay off.

The over-disclosure problem works in your favor

Expect employers to over-label, not under-label, their RIFs as technology displacements. Defense counsel is already warning clients that plaintiffs will scrutinize every reduction in force for an automation nexus, and the $500-per-day Cal/WARN penalty makes under-disclosure the more expensive mistake.

The practical result: borderline restructurings (team consolidations, platform migrations, "efficiency" rounds) will get the "This notice is for a technology displacement" header slapped on them defensively. Your sourcing feed inflates. Some of those notices will tag workers whose roles were only partially automated, which means they come with hands-on experience of the AI system that displaced them. That is a hiring signal, not a disqualifier, for any employer looking for people who have shipped against the exact tool their competitors are now deploying.

Build the EDD intake workflow before January 1, 2027

You have roughly one quarter from signing to operative date. The sourcing teams that win Q1 2027 will have three things in place before the first notice lands.

1. A scraper or watchlist on the EDD technology-displacement index. Individual notices will be posted as they arrive; the quarterly statewide summary rolls up afterward. Both are public.
2. A role-to-skills mapping dictionary. When a notice says "ticket triage and L1 troubleshooting," your pipeline should automatically tag those profiles with Zendesk, Salesforce Service Cloud, Intercom, and the named AI vendor's console.
3. A geography overlay. San Francisco, San Jose, and the broader SF Bay Area are the metros that will produce the bulk of qualifying notices. Pre-build the metro-specific outreach templates now.

Where the engineer notices will cluster

For the Python backend cohort specifically, Refolk's index shows San Francisco, San Jose, and the broader SF Bay Area as the top regions, which is exactly the geography most exposed to coding-agent displacement. If your desk recruits senior backend talent in those metros, SB 951 is going to feed you ex-Amazon, ex-Oracle, and ex-Meta names - the three companies already cited as AI-attributed layoff sources - with employer-attested exposure to generative coding tooling, which is the single hardest signal to confirm from a resume today.

California-only is a feature, for now

SB 951 applies only in California, and that geographic exclusivity is the whole arbitrage. Texas, New York, and Washington have no equivalent disclosure requirement. California accounts for roughly 15% of US tech employment, which means about one in six US AI-driven layoffs will be publicly cataloged with skills tags attached. The rest of the country stays dark.

That window closes the moment another state copies the law, which, given the California Federation of Labor Unions' national footprint, is a 2028 question, not a 2030 one. The sourcing teams who move first get 12 to 18 months of structured competitive intelligence that nobody else can see without manually calling 400 ex-employees to verify what they actually did.

54,966

What to do the day the first notice lands

Treat the first qualifying SB 951 notice as a timed sourcing drill. The 60-day advance notice period means you have a known clock between filing date and the workers' actual last day, which is the exact window where outreach-response rates peak: workers are informed, still employed, and already evaluating options.

• Day 1: Ingest the notice. Parse the four required fields into your ATS as structured tags.
• Day 1 to 7: Build the named-cohort shortlist across LinkedIn, GitHub, and personal sites.
• Day 7 to 30: First-touch outreach, referencing the specific AI system that displaced them. Workers respond to recruiters who already understand their context.
• Day 30 to 60: Second-touch, offers, and backfill against the still-employed peer-company list the EDD summary surfaces.
• Day 60+: Monitor the quarterly EDD roll-up for the next cohort.

The teams who industrialize this workflow by Q2 2027 will have a six-quarter head start on the ones who treat SB 951 as a compliance headline rather than a sourcing feed.

FAQ

When does SB 951 actually take effect?

SB 951 was signed September 30, 2026, and becomes operative on January 1, 2027. The 60-day Cal/WARN advance notice period means the first qualifying technology-displacement notices will cover layoffs with effective dates no earlier than early March 2027, though the notices themselves will post to the EDD starting in January.

What triggers the enhanced disclosure requirement?

Any mass layoff caused "in whole or in substantial part" by an AI system or other automated technology replacing or automating employment positions. The statute does not define "in substantial part," which means defense counsel will advise employers to over-label borderline RIFs rather than risk the $500-per-day Cal/WARN penalty. Expect the feed to be noisier and larger than a strict reading would predict.

Does SB 951 cover independent contractors?

Earlier drafts of the bill defined "worker" to include independent contractors employed for at least six months, a broader category than most WARN-type statutes. The enacted version's final scope on contractors should be confirmed against the signed text, but the direction of travel is clear: SB 951 is designed to catch more workers than Cal/WARN historically has, not fewer.

How do I actually source from the EDD feed on day one?

Monitor the EDD's technology-displacement index for individual notices and the quarterly statewide summary for longitudinal trends. Parse each notice's four required fields (occupation, work location, automated job functions, named AI system) into structured tags, then match those tags against your target pool. Most of the engineering cohort will be identifiable through cross-platform matching across GitHub, LinkedIn, and the open web before their LinkedIn status flips to "open to work."