Managing AI Risks in the Hiring Process

Artificial intelligence tools are being used across an increasingly wide spectrum of applications, each of which presents their own unique risk management challenges.  This Torys memo discusses the use of AI tools in employee recruitment and retention and provides an overview of the regulatory framework and the key risks companies face when using these tools.  This excerpt addresses how government guidance and pending legislation on the use of AI in the workplace provides a risk mitigation roadmap for employers:

The U.S. federal government has issued a number of guidance documents, and several state and local jurisdictions have introduced legislation, that contemplate the use of AI during the hiring process. Other guidance and legislation is directed at regulating employers’ use of AI for its existing workforce, such as for monitoring employees or evaluating them for promotion or termination. Although these items differ in scope and focus, they all reflect similar principles to those enumerated in the AI Bill of Rights and by the Department of Labor in response to the Executive Order. Their requirements are broadly similar and provide employers with a roadmap for addressing current and future developments in this area.

For example, current guidance and legislation typically address the issue of transparency by requiring employers in certain circumstances to provide advance notice to, and possibly obtain consent from, individuals who are subject to algorithmic decision-making tools and inform them as to how information collected by these tools will be used in the employment decision-making process. Employers may want to consider implementing safeguards and best practices for protecting employee data and private information, and limit what information may be gathered and with whom it is shared.

Employers using AI technology in this manner may also consider performing bias audits or impact assessments, put processes in place to address discriminatory results and, where appropriate, make these results available to the government or the public. Another recurring requirement is the provision of alternative, non-AI selection procedures and the presence of processes that allow decisions to be appealed to a human arbiter.

Further, EEOC and ADA Guidance described earlier with respect to hiring decisions demonstrate the importance of providing reasonable accommodations to certain applicants, and employers should be prepared to readily address those requests. Finally, employers may be held accountable for AI tools and software that are developed by, acquired from or administered by third parties, so employers should generally diligence these vendors accordingly.

Among other topics, the memo summarizes pending federal legislation and recently enacted state laws that may impact the use of AI in the workplace.