Last Call! 13 Bills California Employers Should Watch as Governor Newsom's Final Term ...

TEXT START: The California legislature just adjourned and sent a flurry of workplace-related bills related to Governor Newsom to either sign or veto by September 30 – his last go-around before his final term concludes in January.

THE DISSECTION

This is a transition management playbook dressed as employer advisory services. Fisher Phillips—your friendly neighborhood management-side labor law firm—is performing its function with clinical competence: flagging the compliance obligations that arise from a system that knows it's dying but hasn't figured out how to say so.

The operative frame: California is building an increasingly elaborate bureaucratic infrastructure around the margins of AI displacement. Warning notices. Disclosure requirements. Human rubber-stamp requirements. Surveillance restrictions. These are the legal equivalents of rearranging deck chairs on a vessel that has already struck the iceberg—the passengers are being offered better signage to the lifeboats.

THE CORE FALLACY

The "No Robo Bosses Act" (SB 947) is the most instructive example of the fallacy. The bill prohibits employers from relying solely on an Automated Decision System for disciplinary or termination decisions. Notice the surgical precision of the word "solely."

This is a legal fiction. In practice:

• An algorithm generates the recommendation
• A human "reviewer" (often a manager with 30 seconds and a quota) signs off
• The worker gets a post-use notice that reads like a terms-of-service agreement
• The outcome is identical to automated termination

This is procedural theater, not worker protection. It creates documentary evidence that a human was involved, satisfies the letter of the law, and changes nothing about the mechanism of displacement. The human in the loop is a compliance artifact, not a meaningful check.

HIDDEN ASSUMPTIONS

The entire legislative package smuggles in three assumptions that the Discontinuity Thesis renders untenable:

1. Notice and process preserve economic viability. SB 951's WARN Act expansion for AI layoffs assumes that knowing six months in advance that your job is being eliminated by an algorithm somehow maintains your economic participation. It doesn't. It just makes the disorder more scheduled.

2. Human oversight is a meaningful constraint. The surveillance restrictions (AB 1883 banning emotional state inference, neural data collection) assume that how AI is used can be regulated separately from whether AI replaces humans. But the competitive logic doesn't care about your privacy rights once your position has been automated out of existence.

3. Gradualism is the operative timeline. The bills stagger implementation across 2026-2028, as if the transition will be smooth enough to accommodate legislative and employer adaptation windows. This assumes the lag defenses are stronger than the displacement velocity.

SOCIAL FUNCTION

Classification: Transition Management Theater / Legal Copium

This package serves three functions:

1. For capital: Provides a compliance checklist that lets firms continue AI deployment with legal cover. "We followed the SB 947 human-review requirement" is now a defensible position even when the human review is vestigial.

2. For labor: Produces the feeling of protection without the substance. Workers get more paperwork, longer notice periods, and the comforting illusion that the state is watching. This is ideological anesthetic.

3. For the political class: Enables politicians to announce they're "addressing AI in the workplace" while presiding over the largest displacement event in the history of labor. Newsom gets to posture as a tech-forward progressive who also protects workers—a narrative with zero connection to the structural mechanics underneath.

THE VERDICT

California is performing institutional triage on a patient in terminal decline. The bills represent increasingly sophisticated responses to structural collapse, but they are responses within the system that cannot save the system.

The specific mechanics:

• SB 947 won't slow AI-driven termination because "human reviewer" is a rubber stamp, not a meaningful constraint.
• SB 951's WARN expansion acknowledges displacement but doesn't prevent it. Knowing your job is being eliminated by a model instead of a manager doesn't change the outcome.
• AB 1883's neural data restrictions are addressing the margins while the core displacement proceeds. You can protect workers' emotional privacy while they lose their economic purpose.

The real signal: The legislative volume itself is the diagnosis. When a jurisdiction floods its code with AI-related employment law, it's an institutional admission that the old framework is breaking down. The law is building walls against a flood—the walls are better engineered than before, but they're still walls against a flood.

LAG DEFENSE ASSESSMENT

These bills represent legitimate but insufficient lag defenses:

Bottom line: California is spending considerable political and legal capital on measures that will produce compliance costs for employers and procedural comfort for workers—while the fundamental circuit (mass employment → wages → consumption) continues its structural degradation.

THE UNCOMFORTABLE IMPLICATION

The Discontinuity Thesis doesn't predict whether these laws pass. It predicts that they cannot, in aggregate, reverse the displacement dynamic they're responding to.

California can mandate human-in-the-loop reviews, expand WARN notices, restrict surveillance, and add protected categories until the statute books look like a labor lawyer's fever dream. The productive participation problem persists because it's a mathematical problem, not a regulatory problem.

The transition will happen. These laws make it marginally more orderly, which has genuine value for individual workers in the lag window.

But they cannot save the system.

Fisher Phillips knows this. Their job is to help clients navigate the transition, not to pretend the destination is negotiable. The advisory is accurate as far as it goes—it just doesn't go where it matters most.