California’s “No Robo Bosses” Bill Heads to Governor Newsom

California’s “No Robo Bosses” Bill Heads to Governor Newsom: What Employers Need to Know

New legislation targeting the role of artificial intelligence in workforce management is now on Governor Gavin Newsom’s desk.

Senate Bill 947 — the “No Robo Bosses Act of 2026” — would prohibit California employers from using an automated decision system (ADS) as the sole basis for disciplining or terminating an employee. Where such technology plays a primary role in the outcome, a human decision-maker would have to independently corroborate the result before any adverse action takes effect. Workers would also gain the right to written notice that an ADS was used, and to a description of which of their own data factored into the analysis.

The Governor has until September 30, 2026, to sign or veto the bill. If signed, the law would take effect July 1, 2027.

What SB 947 Would Do

The bill is a revised version of SB 7, the original “No Robo Bosses Act,” which Governor Newsom vetoed in October 2025 as overly broad and unfocused — particularly around advance-notice requirements that could have swept in routine workplace tools. The 2026 rewrite narrows the focus to discipline and termination, replaces pre-use notice with post-use notice, and keeps a human in the loop.

Key provisions, if enacted:

1. No “robo-only” discipline or termination. An employer could not rely solely on an ADS when making a disciplinary or termination decision.
2. Human corroboration when ADS is primary. If the employer primarily relies on ADS output, a human reviewer would have to corroborate the decision using the underlying data or other supporting information — such as evaluations, personnel files, work product, peer reviews, or witness interviews. If the output cannot be corroborated, or the reviewer finds it inaccurate, incomplete, or misleading, it could not be used for the decision.
3. Written post-use notice. When an employer primarily relied on an ADS, the affected employee would receive a plain-language written notice at the time the decision is communicated. The notice would state that an ADS was primarily used, that a human reviewed and corroborated the output, how to contact the reviewer, how to request a description of the employee’s own data, and that retaliation is prohibited.
4. Employee data-description right. Upon request, the employer would have to provide a meaningful, objective description of the employee’s own data used by the ADS. Personal information about other people would need to be anonymized.
5. Additional ADS bans. Employers could not use an ADS to violate labor, safety, employment, or civil rights laws; to infer a worker’s protected status under the Fair Employment and Housing Act; or to predict and take adverse action against a worker for exercising legal rights.
6. Anti-retaliation and enforcement. Retaliation for asserting rights under the law would be prohibited. The Labor Commissioner and public prosecutors could enforce the statute. Civil penalties would include $500 per violation, with additional remedies available in court. Once ADS use in a disciplinary decision is shown, the employer would bear the burden of proving it did not primarily rely on the system — or that it complied with the corroboration and notice rules.

The definition of “ADS” is broad: any computational process derived from machine learning, statistical modeling, data analytics, or artificial intelligence that issues a simplified output (a score, classification, or recommendation) used to assist or replace human discretionary decision-making and that materially impacts people. Spam filters, firewalls, antivirus tools, identity-access systems, calculators, and raw databases are excluded. The bill would cover private and public employers, as well as labor contractors, and would apply to charter cities. A valid collective bargaining agreement could waive the law only if it expressly does so and includes specified worker protections.

Related California Measures — and Federal Pushback

SB 947 is not the only workplace-AI bill to leave the Legislature this session. Lawmakers also advanced separate measures that would limit employer use of tools that collect neural data or infer emotional state, and that would require advance warning when job eliminations result from the adoption of automated technologies. Those bills, like SB 947, still depend on the Governor’s action.

California would be among the first jurisdictions to put this kind of guardrail around algorithmic management. That state-level activity is colliding with a different approach in Washington. The current federal administration has signaled an intent to limit states’ ability to write their own AI rules — through proposed federal legislation, Department of Justice litigation challenging certain state AI laws, and the possibility of tying discretionary federal funds to a state’s regulatory climate. None of that has wiped SB 947 off the books. It does mean multi-state employers should plan for a period of overlapping, and possibly conflicting, requirements rather than a single national standard.

What Employers Should Do Now

SB 947 is not law yet. Treating it as a planning document — not a compliance checklist — is the right posture until the Governor acts. If you operate in California or use automated tools that feed discipline, performance management, productivity scoring, or termination decisions, these steps will put you in a better position either way:

1. Inventory the tools. List every system that scores, ranks, flags, recommends, or otherwise influences discipline, performance ratings, attendance enforcement, productivity monitoring, or separation decisions. Include vendor platforms, not just tools built in-house.
2. Map “sole” versus “primary” use. For each tool, document whether a human currently makes the final call, what information that person reviews, and whether the system’s output is effectively driving the outcome. “The manager clicked approve” is not the same as independent corroboration.
3. Separate the score from the file. Build a review path that can stand on evaluations, personnel records, work product, and interviews — the kinds of sources the bill treats as corroborating information — rather than on a dashboard score alone.
4. Draft the notice you would need. Even if the bill is vetoed, a plain-language explanation of when automated tools contributed to an adverse action is a useful discipline for managers and a useful record if a decision is later challenged.
5. Watch the September 30 deadline — and July 1, 2027. If the bill is signed, you would have until mid-2027 to adjust policies, vendor contracts, manager training, and documentation practices. If it is vetoed, California’s existing privacy rules on automated decision-making technology may still reach some of the same tools.
6. Do not assume a federal override will arrive in time. Federal preemption efforts are real, but they are unfinished. California employees and California worksites should be planned for under California rules unless and until a court or Congress actually displaces them.

This is a developing area. The right process for a 12-person shop using a basic attendance flag is not the same as the process for a multi-state employer running AI-supported performance management. The facts of the tool, the job, and the decision still matter.