What's Happening?
California has passed Senate Bill (SB) 947, known as the 'No Robo Bosses Act,' which imposes new restrictions on how employers can use Automated Decision Systems (ADS) in the workplace, particularly concerning disciplinary and termination decisions. Effective
July 1, 2027, the law prohibits employers from relying solely on ADS for such critical employment actions. If an employer primarily relies on an AI-powered tool's output for discipline or termination, a human must corroborate the decision using the data collected by the ADS or other relevant supporting information like supervisory evaluations or personnel files. The law broadly defines ADS to include computational processes derived from machine learning, statistical modeling, data analytics, or artificial intelligence that produce simplified outputs like scores or recommendations materially impacting individuals. It also mandates that employers provide affected employees with a written 'postuse' notice if an ADS was primarily used in a disciplinary or termination decision, detailing that a human reviewed and corroborated the output, and providing contact information for questions.
Why It's Important?
This legislation marks a significant step in regulating artificial intelligence in the workplace, aiming to prevent over-reliance on automated systems for high-stakes employment decisions. It underscores a growing concern about algorithmic bias and the potential for AI to make unfair or unexplainable decisions without human oversight. For employers, the law introduces new compliance burdens, requiring them to audit their existing AI tools, establish clear human review processes, and develop new notification protocols. The undefined term 'primarily relies' is likely to be a point of contention and litigation, creating uncertainty for businesses. Employees gain new rights to transparency and human review in disciplinary and termination processes, potentially leading to more equitable outcomes and reducing the risk of automated discrimination. This law could also influence other states and federal lawmakers to consider similar regulations, setting a precedent for AI governance in employment.
What's Next?
Employers in California need to begin preparing for compliance well in advance of the July 1, 2027, effective date. This includes inventorying all automated decision systems currently in use, especially those that generate scores, rankings, or recommendations related to discipline or termination. Companies will need to assess how these tools are integrated into their decision-making processes and ensure that robust human review mechanisms are in place. Developing standardized 'postuse' notice templates and training managers on the new requirements will also be crucial. The law does not create a private right of action but allows the labor commissioner and public prosecutors to enforce it, with potential civil penalties and injunctive relief. Employers may also face claims under California’s Private Attorneys General Act (PAGA). Future litigation is expected to clarify ambiguities, particularly around the definition of 'primarily relies' on ADS output.
Beyond the Headlines
The 'No Robo Bosses Act' reflects a broader societal debate about the role of AI in human decision-making and accountability. It highlights the ethical imperative to ensure that technological advancements do not erode fundamental worker protections or introduce new forms of discrimination. The law's focus on human corroboration suggests a hybrid model where AI can assist but not fully replace human judgment in critical areas. This could lead to a re-evaluation of AI development in HR technology, pushing vendors to design systems that are more transparent, auditable, and supportive of human oversight rather than fully autonomous. The legislation also touches upon the evolving concept of workplace privacy and data access rights, as employees can request descriptions of their data used by ADS. This could foster a more transparent and trust-based relationship between employers and employees regarding data usage and AI deployment.
