# Mental Health AI Safety Claims Must Preserve Temporal Evidence

arXiv: https://arxiv.org/abs/2605.08827 · HTML: https://mlsanity.com/paper/2605.08827/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 7 of its 24 references that are in the map.
- First posted on arXiv 9 May 2026.

## Builds on (most cited first, 7 of 7)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [CounselBench: A Large-Scale Expert Evaluation and Adversarial Benchmarking of Large Language Models in Mental Health Question Answering](https://mlsanity.com/paper/2506.08584/) (2025)
- [MindEval: Benchmarking Language Models on Multi-turn Mental Health Support](https://mlsanity.com/paper/2511.18491/) (2025)
- [Building Trust in Mental Health Chatbots: Safety Metrics and LLM-Based Evaluation Tools](https://mlsanity.com/paper/2408.04650/) (2024)
- [MHSafeEval: Role-Aware Interaction-Level Evaluation of Mental Health Safety in Large Language Models](https://mlsanity.com/paper/2604.17730/) (2026)
- [MHDash: An Online Platform for Benchmarking Mental Health-Aware AI Assistants](https://mlsanity.com/paper/2602.00353/) (2026)
- [EchoGuard: An Agentic Framework with Knowledge-Graph Memory for Detecting Manipulative Communication in Longitudinal Dialogue](https://mlsanity.com/paper/2603.04815/) (2026)

## Built on by (most cited first, 1 of 1)

- [Quantifying the Relationship Between Clinical Safety and Environmental Impact in Therapeutic LLMs](https://mlsanity.com/paper/2608.11830/) (2026)

## Abstract (arXiv)

The safety of mental health AI is often judged at the wrong temporal scale. Current evaluations typically score isolated responses, endpoint outcomes, or aggregate dialogue quality, while clinically consequential failures may arise from the order and accumulation of interactions themselves, including delayed escalation, repeated reinforcement, dependency formation, failed repair, and gradual deterioration across turns. This paper argues that this mismatch is not merely a limitation of evaluation coverage but a source of invalid safety conclusions. We introduce Temporal Safety Non-Identifiability, a formal account of why safety properties that depend on sequence, timing, accumulation, or recovery cannot be certified by protocols that discard those features. From this formalization, we develop SCOPE (Safety Claims Over Preserved Evidence) as a general principle for aligning safety claims with the evidence an evaluation actually retains, and instantiate it as SCOPE-MH, a mental-health instantiation of this reporting standard. We operationalize SCOPE-MH through a proof-of-concept on the AnnoMI dataset of expert-annotated motivational interviewing conversations, which reveals mechanisms of failure that per-turn behavior scoring does not represent. We propose SCOPE-MH as a diagnostic complement to existing evaluation infrastructure and argue that evaluation preserving temporal evidence is necessary, not optional, for safety-critical mental health AI deployment.