MindGuard: Guardrail Classifiers for Multi-Turn Mental Health Support
This addresses safety failures in mental health AI applications by providing more precise safeguards, though it is incremental as it builds on existing classifier methods with clinical adaptations.
The paper tackles the problem of ensuring clinical appropriateness in large language models used for mental health support by introducing MindGuard, a family of lightweight safety classifiers trained on a clinically grounded risk taxonomy and synthetic dialogues, which reduce false positives and lower harmful engagement rates in adversarial interactions compared to general-purpose safeguards.
Large language models are increasingly used for mental health support, yet their conversational coherence alone does not ensure clinical appropriateness. Existing general-purpose safeguards often fail to distinguish between therapeutic disclosures and genuine clinical crises, leading to safety failures. To address this gap, we introduce a clinically grounded risk taxonomy, developed in collaboration with PhD-level psychologists, that identifies actionable harm (e.g., self-harm and harm to others) while preserving space for safe, non-crisis therapeutic content. We release MindGuard-testset, a dataset of real-world multi-turn conversations annotated at the turn level by clinical experts. Using synthetic dialogues generated via a controlled two-agent setup, we train MindGuard, a family of lightweight safety classifiers (with 4B and 8B parameters). Our classifiers reduce false positives at high-recall operating points and, when paired with clinician language models, help achieve lower attack success and harmful engagement rates in adversarial multi-turn interactions compared to general-purpose safeguards. We release all models and human evaluation data.