California is among the first states to propose expressly regulating employers’ use of algorithms and artificial intelligence. In a virtual public meeting on March 25, 2022, the California Fair Employment and Housing Council discussed proposed regulatory changes that would address the use of artificial intelligence by employers and third parties in employment practices. While the proposed regulations remain a work in progress, they provide insight into how policymakers are approaching these issues — and they could prove influential for other states (and even, potentially, the federal government) considering their own regulations in this space.
A Closer Look: AI, Algorithms, and Job-Related Decisions
The proposed rule, which was released on March 15, 2022, would clarify that current California non-discrimination law prohibits an employer’s use of “qualification standards, job tests, automated decision systems or ‘other selection criteria which eliminate or tend to exclude a candidate or an employee or a category of candidates or employees’ on the basis of characteristics protected by law, unless the selection criteria prove to be ‘linked to employment” and “compatible with the needs of the company”.
Specifically, the proposed regulations:
- Define “automated decision system” broadly as any “computational process…or any other data process or artificial intelligence technique…” and specifically includes algorithms that:
- The screen resumes for terms or patterns.
- Use facial or voice recognition to analyze phrases or word choices.
- Use game-based tests that include puzzles or questions to perform predictive assessments or measure characteristics such as dexterity, reaction time, or other physical or mental abilities or characteristics.
- Use online tests to measure personality traits, aptitudes, cognitive abilities, and cultural fit.
- Expand the definition of “employment agency” to include any person who provides automated decision-making systems or services involving the administration or use of such systems on behalf of an employer.
- Extend record-keeping requirements from two to four years and make explicit the requirement to retain “all machine learning data.”
- Expand the aiding and abetting provision by adding that unlawful aiding, soliciting and abetting includes the “advertising, selling, providing or using” a selection tool that restricts , unlawfully screens out or discriminates against applicants or employees based on protected characteristics.
During the March 25 meeting, the subcommittee that drafted the proposed regulatory revisions provided an overview of the drafting process and solicited input from Board members and public comment. The discussion suggested that the proposed settlements are not intended to create new causes of action or defence. Rather, they are intended to clarify the application of current law to employers’ current and planned use of emerging technologies to make employment-related decisions.
Several commenters have asked about the proposed regulations’ use of highly technical terms, which may be unfamiliar to those outside of the tech industry. Others asked if the subcommittee had reviewed laws in other jurisdictions or sought input from subject matter experts. The most prominent question, however, concerned the timing of the next steps. In response to these questions, the Commission clarified that the 45-day comment period does not have begin with the March 15, 2022 publication of the proposed regulations. The Council is still “working” on the proposed regulations and continues to invite the informal public to comment on the draft.
The proposed settlement follows the Board’s April 30, 2021 virtual civil rights hearing on the use of algorithms in employment, housing, loans and healthcare. The Council has prioritized the review of employment-related regulations, but proposed regulations relating to the use of algorithms in housing are expected to follow.
A sign of things to come?
While California is among the first to propose expressly regulating employers’ use of algorithms, other states — and the federal government — are likely close behind. In the Equal Employment Opportunity Commission (EEOC) press release dated March 28, 2022 regarding its fiscal year 2021 performance report and fiscal year 2023 budget justification, the EEOC announced that its proposed budget will help advance three initiatives launched in 2021, one of which is “an initiative to ensure that job-related artificial intelligence and algorithmic decision-making tools are compliant with federal civil rights laws Simply put, there is no doubt that employers will face increased regulation of the use of automated decision-making tools in the near future, in California and beyond.