California is seeking to regulate the use of AI in mental health treatment, addressing a gap that has opened faster than either clinical governance or licensing frameworks: AI chatbots are already being used for intake, triage, therapy-adjacent conversation and crisis signposting, in some cases without clinician oversight, and the state's proposal would impose requirements on how those systems operate rather than banning them. The healthcare relevance is regulatory rather than technical — it is the first US state action aimed specifically at the intersection of a clinical duty of care and an unlicensed automated agent, and it follows the same pattern as the AI incident-response gap and the rogue-agent incidents that defined this week's AI-safety coverage. For health providers, the practical consequence is that any AI function touching patient mental health data in California becomes a compliance surface with documentation and oversight obligations, and telehealth and digital-mental-health vendors operating nationally will need to decide whether to build to California's standard or run separate configurations.

• Source article