Joseph O’Keefe Speaks With Lexology PRO About How AI Tools Could Trigger Genetic Discrimination Claims

Labor and employment partner Joseph O’Keefe spoke with Lexology PRO about how employers’ use of AI intake tools, chatbots, and third-party platforms in recruitment, occupational health, benefits, leave, accommodations, and employee support functions could create exposure under the Genetic Information Nondiscrimination Act of 2008 (GINA).

O’Keefe explained that GINA distinguishes protected genetic information from medical information governed by the Americans with Disabilities Act (ADA), noting that an employee’s own manifested disease or disorder is generally ADA medical information, while a relative’s manifestation of a disease or disorder is genetic information under GINA.

“An employee’s own manifested disease or disorder, something the employee actually has, is generally medical information under the ADA, not genetic information under GINA. But a relative’s manifestation of disease or disorder is genetic information under GINA,” O’Keefe said.

O’Keefe said there is potential GINA exposure “basically anywhere a chatbot or AI tool is asking health-related questions of employees or applicants.” He identified high-risk prompts that ask about family health conditions, request family health trees, seek medical records related to a condition, or gather health information to personalize wellness plans.

He also noted that generative AI tools can create risk through dynamic follow-up questions, such as when an employee mentions a family history of cancer and the tool asks about genetic screening. “Medically logical doesn’t mean legally permissible,” O’Keefe said.

To mitigate risk, O’Keefe encouraged employers to inventory AI and chatbot tools that touch on health-related topics, review prompts, forms, upload requests, generated follow-ups, training data, and knowledge bases, and require vendors to agree to GINA-specific compliance representations, indemnity covering GINA and analogous state claims, and guardrails that prevent AI tools from generating genetic-information follow-up questions.