OpenAI published a blog post on 25 November 2025 outlining its litigation strategy for mental health-related claims [source]. The post describes how the company intends to respond to lawsuits alleging harm from interactions with its AI models in mental health contexts.

The company states it will "vigorously defend" against claims it considers meritless, while acknowledging that its models are not designed to provide clinical mental health care. OpenAI says it has implemented safety measures including crisis resource signposting and refusal behaviours for certain prompts, but does not specify which prompts trigger these interventions.

The post does not disclose whether OpenAI has received formal legal complaints or demand letters related to mental health harms. It references "increasing public discussion" of AI safety in sensitive domains but provides no case studies or incident data.

OpenAI emphasises that its terms of service prohibit use of its models for medical diagnosis or treatment, and that users are warned against relying on outputs for health decisions. The company says it will seek dismissal of claims where plaintiffs cannot demonstrate the model was used outside its intended scope.

The post arrives amid broader scrutiny of generative AI in healthcare settings. Regulators in the EU and UK have flagged mental health applications as high-risk, and several academic studies have documented instances where chatbots provided harmful advice in simulated crisis scenarios.

OpenAI did not announce changes to its models, safety filters, or user-facing disclosures in the post. The company says it will continue to refine its approach based on "emerging case law and regulatory guidance."