Ling-Yi Tsai is a veteran at the intersection of human resources and advanced technology, having spent decades guiding global organizations through the complex integration of AI and talent analytics. As a specialist in recruitment and onboarding systems, she offers a unique perspective on the ethical and legal friction points that arise when data-driven efficiency meets worker protections. With California’s latest legislative shift toward regulating “automated bosses,” her insights provide a roadmap for compliance and human-centric management.
The evolution from the initial legislative attempts to the refined protections in the current law suggests a shifting political landscape—how do these specific changes impact the way businesses must now view their AI tools?
The journey from the vetoed Senate Bill 7 in October 2025 to the successful signing of SB 947 this past September reflects a more focused approach to curbing algorithmic bias. Governor Newsom previously expressed concerns that the earlier version was too broad, potentially hampering innocuous tools like spam filters or antivirus software. By refining the “No Robo Bosses Act,” lawmakers have specifically targeted consequential actions like employee discipline and termination, moving away from unfocused notification requirements. This means that as we move through 2026, businesses must scrutinize their machine learning and statistical modeling tools with a much higher degree of precision. It is a clear signal that while innovation is encouraged, the weight of a person’s livelihood cannot be left to a script that lacks human discretion.
With the new requirement that human reviewers must corroborate automated recommendations, what does a robust and legally defensible review process actually look like on the ground?
A “rubber-stamp” approval is no longer sufficient; the law demands a meaningful, independent verification of the algorithm’s output. To be compliant by the July 1, 2027, deadline, a human reviewer must actively dig into managerial evaluations, personnel records, and actual employee work products to see if they align with the AI’s score or classification. Imagine a scenario where a system flags a warehouse worker for a drop in productivity; the human reviewer is now obligated to look for external factors, perhaps conducting witness interviews or reviewing peer feedback, to ensure the data isn’t misleading. If the human finds that the automated system’s output is inaccurate or incomplete, that data point is legally barred from being the basis for any adverse action. This process injects a necessary layer of empathy and context, ensuring that the cold logic of a computer is tempered by the nuanced understanding of a real manager.
In what ways do the new transparency requirements change the psychological contract and communication dynamic between an employer and their workforce?
The transparency mandates under SB 947 are designed to eliminate the “black box” feeling that often accompanies modern management, where employees feel they are being judged by an invisible, uncaring force. When an employer primarily relies on an automated decision system for discipline, they are now required to provide a written notice that explicitly confirms a human has reviewed and corroborated the findings. This notice must also include a pathway for the employee to contact a real person for additional information and provide them the right to request a meaningful description of the employee data used by the system. It shifts the power dynamic from one of suspicion to one of accountability, allowing workers to feel that their rights to fair employment are being respected even in a tech-driven environment. There is a profound emotional difference for an employee when they know their performance isn’t just a number on a dashboard, but a documented record that a human has carefully considered.
As we look toward the implementation date next year, what immediate actions should HR leaders take to audit their current technologies and ensure they aren’t inadvertently violating these new standards?
The first step for any HR department in 2026 is to perform a comprehensive inventory of every computational process that assists or replaces human discretionary decision-making. You have to look beyond simple databases and identify any tool using machine learning or data analytics to generate scores, classifications, or recommendations for staff. It’s critical to remember that this law also prohibits using these systems to infer an employee’s protected status or to take action against workers for exercising legal rights, which requires a deep dive into the underlying code and data sets of your vendors. Organizations should also pay close attention to the intersection of this law with other recent measures, such as those regarding AI-powered surveillance and mass layoff disclosures. By building a documented trail of human corroboration now, companies can avoid the legal pitfalls that SHRM and other industry groups have highlighted during the legislative process.
What is your forecast for algorithmic management?
I predict that the “California effect” will lead to a nationwide standard where the concept of the “Robo Boss” is replaced by a “Co-Pilot” model of management. As we move from 2026 into the next few years, the legal expectation of human oversight will force AI developers to create more explainable and transparent models that prioritize ethical outcomes over raw efficiency. We will see a significant shift in HR training, where algorithmic literacy becomes a core competency for managers who must be able to challenge and verify automated outputs. Ultimately, the companies that thrive will be those that view these regulations not as a burden, but as an opportunity to rebuild trust with a workforce that is increasingly wary of automation. The future isn’t about removing AI from the workplace, but about ensuring it serves as a tool for human empowerment rather than a replacement for human judgment.