Over the past year, we have seen the various agencies and branches of government issue a slew of rules, guidance, and directives over the use (and misuse) of Artificial Intelligence (AI). In May, the EEOC issued guidance on the use of automated systems in employment decisions. November 2023 brought the White House’s AI Executive Order, directing federal agencies to develop further guidelines around areas of privacy, national security, consumer protection, cybersecurity, and intellectual property. We have discussed both the EEOC and White House developments in detail.
The U.S. Securities and Exchange Commission (SEC) has also weighed in, proposing a new rule to address the risks associated with the use of predictive data analytics by broker-dealers and investment advisers. The proposed rule sets forth a number of requirements for broker-dealers and investment advisors to avoid potential conflicts between firms and the investing public.
These measures are only a first step, and 2024 is primed for a patchwork of state AI regulation.
A recent bill that failed to advance through the California Legislature is an instructive blueprint for state and local laws to come. Assembly Bill 331, which died in appropriations last May, proposed requirements for users of AI to perform an impact assessment analyzing — among other things — the potential adverse impacts on legally protected characteristics, such as sex, race, and age, to name a few.
Under the bill, failure to send the assessment to the state’s Civil Rights Department within 60 days of completion would result in a potential administrative fine not to exceed $10,000. The bill’s other conditions included: a notice requirement to persons who are the subject of the technology’s consequential decisions; a prohibition of the use of AI contributing to algorithmic discrimination; development of a statement of intended uses and limitations; and the establishment, implementation, and maintenance of a safeguarding program designed to prevent algorithmic discrimination. These requirements and others would apply to employers if adopted.
Though unsuccessful in its first iteration, the California bill saw considerable support in committee and is likely to return soon — along with others like it at the state and local levels.
Without a uniform system of regulation from Congress, companies can expect states and municipalities to lead the way.
The resulting variance will present compliance challenges, particularly for those operating across state borders. We will continue to monitor these developments as employers consider their use of AI technology.