What Should Clinicians Disclose to Clients Before Incorporating AI Into Assessments or Documentation?

What should clinicians disclose to clients before incorporating AI into assessments or documentation?

Clients should be informed from the outset, even during session scheduling, about any intended use of AI, rather than being told at the last minute that a session is being recorded or analyzed. This disclosure should cover the specific nature of the AI use (e.g., whether it is simply correcting grammar or actually generating diagnostic interpretations or treatment plans), along with an honest discussion of the associated benefits and risks. Wherever possible, this rationale should be grounded in available research rather than speculation, so that clients can make a genuinely informed choice.

Respecting informed consent also means offering alternatives. If a client objects to being recorded or has concerns rooted in trauma history, involvement in legal proceedings, or fear of how their information might be used by authorities, the clinician should honor that preference by using traditional note-taking methods instead. Even in circumstances where consent is not legally mandated, such as certain involuntary or court-mandated contexts, providing clients with informed notice about AI use, explaining what is being done, why, and what recourse they have, remains the ethically preferable approach.

This Ask the Expert is an edited excerpt from the course, 'Ethics of AI for Clinical Assessments and Documentation,' presented by Allan Barsky, JD, MSW, PhD.