California Workplace AI: What Stalled Legislation Reveals About Employer Disclosure, Documentation, and Oversight Obligations

Although California’s Senate Bill 7 did not become law and California’s Assembly Bill 1018 is currently inactive, both bills preview the types of disclosure, documentation, and oversight obligations that California lawmakers have been considering in connection with employers’ use of automated tools in employment decisions.

Many employers are already using automated tools to screen applicants, evaluate performance, and support hiring and disciplinary decisions. In some cases, however, the people using those tools do not understand them well. For example, employers may not know what inputs the tool relies on, how it generates outputs, or how the tool applies those outputs in “decision-making.” That gap creates compliance risk under existing law because the employer may not be able to explain or substantiate how a particular employment decision was made.

Since October 1, 2025, California employers have been operating under the state’s Civil Rights Council’s automated-decision-system regulations, which clarify how existing anti-discrimination law applies when employers use artificial intelligence, algorithms, and other automated tools in employment decisions. Those regulations reinforce that employers remain responsible for the use and outcomes of these tools.

Against that backdrop, the California Legislature’s two recent AI-related proposals are best understood as addressing a related issue: not only whether an automated system produces a discriminatory result but whether the employer can demonstrate how that system is used, what it relies on, and what controls govern its use.

First, the now-failed Senate Bill 7, sometimes referred to as the “No Robo Bosses Act,” would have directly regulated employer use of automated decision systems in employment by requiring notice to applicants and employees, imposing recordkeeping obligations, and restricting reliance on automated outputs in making certain employment decisions.

Second, the now-inactive Assembly Bill 1018 would have taken a broader approach by imposing requirements on both developers and users of automated decision systems used in consequential decisions, including employment decisions, such as impact assessments, disclosures, governance controls, and opportunities for human review.

Although Senate Bill 7 did not become law and Assembly Bill 1018 has not been enacted, both reflect a consistent focus on disclosure, documentation, and oversight.

For employers, the significance of these proposals is not that they proposed a new theory of liability but that they identify a potential emerging consensus on compliance expectations. The key questions are practical: can the employer explain what tools are being used, what those tools do, what data they rely on, and how their outputs are incorporated into employment decisions? Where the answer is unclear, the employer may face exposure because the employer cannot substantiate how the “employer” (through an AI tool) made an employment-related decision.

These abstract issues can arise in concrete ways. If an automated screening tool excludes a category of applicants, the employer should be able to explain what criteria drove that result and why those criteria are job-related. If a productivity or monitoring tool evaluates employee conduct, the employer should be able to show that the tool’s outputs are reviewed in context and not applied mechanically. If a hiring or disciplinary decision relies on a score, ranking, or recommendation generated by a system, the employer should be able to explain how that output was generated and whether it was independently evaluated before action was taken. These documentation and oversight issues directly affect the employer’s ability to explain (and therefore defend) its decisions under existing anti-discrimination law.

Viewed through that lens, these recent legislative proposals point toward a framework centered on three core requirements:

1. Disclosure. Employers should expect increasing pressure to disclose when automated tools are used in ways that materially affect hiring, discipline, promotion, compensation, or other employment decisions.
2. Documentation. Employers should be able to identify what systems are in use, what those systems are designed to do, what data they rely on, and how their outputs are determined. This includes maintaining records sufficient to explain how a given tool factored into a specific employment decision.
3. Oversight. Employers should ensure that automated outputs are subject to meaningful human review, including the ability to evaluate and, where appropriate, override system-generated results rather than relying on them as determinative.

Practically, many organizations are not structured around these requirements. Automated tools are often deployed across different functions, including recruiting, performance management, scheduling, and operations — without centralized visibility. Vendors may control key aspects of system functionality, while employer-side documentation about how the tool operates and how it is used remains limited. Managers may rely on outputs without understanding their generation or any system limitations. That lack of coordination can compound compliance risk.

Addressing that risk does not require waiting for new legislation. Employers should conduct an internal inventory of automated tools used in employment decision-making, identify where those tools influence specific decisions, and ensure that their use is documented in a way that can be explained and supported. Employers should also assess whether there is meaningful human review before taking adverse action and whether vendor relationships provide sufficient information about how systems function, including any limitations or assumptions built into them. Finally, employers should evaluate whether they could clearly articulate, if required, how a particular tool was used in a particular employment decision.

The fact that Senate Bill 7 did not become law and Assembly Bill 1018 is currently inactive does not diminish their relevance. They provide a clear indication of where California is focusing its attention: not only on outcomes but on whether employers can demonstrate control over the tools they use. Employers that align their practices around disclosure, documentation, and oversight will be better positioned to meet those expectations under existing law and under whatever framework comes next.