Beyond Simulations: What 20,000 Real Conversations Reveal About Mental Health AI Safety

Large language models (LLMs) are increasingly used for mental health support, yet existing safety evaluations rely primarily on small, simulation-based test sets that have an unknown relationship to the linguistic distribution of real usage. In this study, we present replications of four published safety test sets targeting suicide risk assessment, harmful content generation, refusal robustness, and adversarial jailbreaks for a leading frontier generic AI model alongside an AI purpose built for mental health support. We then propose and conduct an ecological audit on over 20,000 real-world user conversations with the purpose-built AI designed with layered suicide and non-suicidal self-injury (NSSI) safeguards to compare test set performance to real world performance. While the purpose-built AI was significantly less likely than general-purpose LLMs to produce enabling or harmful content across suicide/NSSI (.4-11.27% vs 29.0-54.4%), eating disorder (8.4% vs 54.0%), and substance use (9.9% vs 45.0%) benchmark prompts, test set failure rates for suicide/NSSI were far higher than in real-world deployment. Clinician review of flagged conversations from the ecological audit identified zero cases of suicide risk that failed to receive crisis resources. Across all 20,000 conversations, three mentions of NSSI risk (.015%) did not trigger a crisis intervention; among sessions flagged by the LLM judge, this corresponds to an end-to-end system false negative rate of .38%, providing a lower bound on real-world safety failures. These findings support a shift toward continuous, deployment-relevant safety assurance for AI mental-health systems rather than limited set benchmark certification.

Adversarial Training for Failure-Sensitive User Simulation in Mental Health Dialogue Optimization

Realistic user simulation is crucial for training and evaluating task-oriented dialogue (TOD) systems, yet creating simulators that accurately replicate human behavior remains challenging. A key property of effective simulators is their ability to expose failure modes of the systems they evaluate. We present an adversarial training framework that iteratively improves user simulator realism through a competitive dynamic between a generator (user simulator) and a discriminator. Applied to mental health support chatbots, our approach demonstrates that fine-tuned simulators dramatically outperform zero-shot base models at surfacing system issues, and adversarial training further enhances diversity, distributional alignment, and predictive validity. The resulting simulator achieves a strong correlation between simulated and real failure occurrence rates across diverse chatbot configurations while maintaining low distributional divergence of failure modes. Discriminator accuracy decreases drastically after three adversarial iterations, suggesting improved realism. These results provide evidence that adversarial training is a promising approach for creating realistic user simulators in mental health support TOD domains, enabling rapid, reliable, and cost-effective system evaluation before deployment.

Past, Present, and Future: A Survey of The Evolution of Affective Robotics For Well-being

Recent research in affective robots has recognized their potential in supporting human well-being. Due to rapidly developing affective and artificial intelligence technologies, this field of research has undergone explosive expansion and advancement in recent years. In order to develop a deeper understanding of recent advancements, we present a systematic review of the past 10 years of research in affective robotics for wellbeing. In this review, we identify the domains of well-being that have been studied, the methods used to investigate affective robots for well-being, and how these have evolved over time. We also examine the evolution of the multifaceted research topic from three lenses: technical, design, and ethical. Finally, we discuss future opportunities for research based on the gaps we have identified in our review -- proposing pathways to take affective robotics from the past and present to the future. The results of our review are of interest to human-robot interaction and affective computing researchers, as well as clinicians and well-being professionals who may wish to examine and incorporate affective robotics in their practices.