mlsanityData to 14 Oct 2026

One Year Later...The Harms Persist, But So Do We!

arXiv 2606.23884 ↗First posted 22 Jun 2026cs.CL cs.AIcited by 0

On the map: LLM safety and security → Safety alignment, refusal and guardrails

Abstract (arXiv)

General-purpose large language models (LLMs) are increasingly used for mental health-related conversations, yet safety guardrails remain inadequate and inconsistent across clinical conditions. This study evaluates eight proprietary LLMs across 16 DSM-5 conditions using four adversarial attack variants, introducing an eight-dimension harm taxonomy and a multi-dimensional evaluation framework. Results show that safeguards hold reliably only for suicide and self-harm, while conditions such as eating disorders, substance use disorder, and major depressive disorder exhibit failure rates of up to 100%. We argue that ethical design and deployment of these LLMs demand clearly defined harm categories across clinical conditions and implementation of safeguards accordingly. Until such safeguards are in place, these models pose significant risks to vulnerable populations, making their growing integration into publicly available settings (e.g., schools, search engines, and consumer chatbots) are particularly concerning.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 26 of its 57 references that are in the map. First posted on arXiv 22 Jun 2026.

Builds on

26 of 57 references.

Closest ancestors · 2

Same direction, newest first

  1. 2026Differential Harm Propensity in Personalized LLM Agents: The Curious Case of Mental Health Disclosure17 Mar
  2. 2025Towards Safe AI Clinicians: A Comprehensive Study on Large Language Model Jailbreaking in Healthcare27 Jan

Other references · 24

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  3. 2023Jailbroken: How Does LLM Safety Training Fail?cited by 1,651
  4. 2024HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusalcited by 1,429
  5. 2023Towards Understanding Sycophancy in Language Modelscited by 1,040
  6. 2022Red Teaming Language Models with Language Modelscited by 1,026
  7. 2022Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learnedcited by 729
  8. 2023Jailbreaking ChatGPT via Prompt Engineering: An Empirical Studycited by 498
  9. 2023"Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Modelscited by 489
  10. 2024Prometheus 2: An Open Source Language Model Specialized in Evaluating Other Language Modelscited by 446
  11. 2024A StrongREJECT for Empty Jailbreakscited by 385
  12. 2023JudgeLM: Fine-tuned Large Language Models are Scalable Judgescited by 367
12 more
  1. 2023Multi-step Jailbreaking Privacy Attacks on ChatGPTcited by 333
  2. 2023PandaLM: An Automatic Evaluation Benchmark for LLM Instruction Tuning Optimizationcited by 316
  3. 2024SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusalcited by 240
  4. 2023Generative Judge for Evaluating Alignmentcited by 169
  5. 2023MART: Improving LLM Safety with Multi-round Automatic Red-Teamingcited by 163
  6. 2024BadChain: Backdoor Chain-of-Thought Prompting for Large Language Modelscited by 80
  7. 2024The opportunities and risks of large language models in mental healthcited by 54
  8. 2023Benefits and Harms of Large Language Models in Digital Mental Healthcited by 40
  9. 2023LLM-empowered Chatbots for Psychiatrist and Patient Simulation: Application and Evaluationcited by 34
  10. 2024Large Language Models in Mental Health Care: a Scoping Reviewcited by 34
  11. 2026Assessing the Quality of Mental Health Support in LLM Responses through Multi-Attribute Human Evaluationcited by 6
  12. 2025Independent Clinical Evaluation of General-Purpose LLM Responses to Signals of Suicide Riskcited by 4

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025`For Argument's Sake, Show Me How to Harm Myself!': Jailbreaking LLMs in Suicide and Self-Harm Contexts
  2. 2025CARES: Comprehensive Evaluation of Safety and Adversarial Robustness in Medical LLMs
  3. 2026Retrieval-Augmented Generation in LLMs for Mental Health: Quantifying the Incremental Contribution of Retrieval Within a Layered Safety Architecture
  4. 2026MindGuard: Guardrail Classifiers for Multi-Turn Mental Health Support
  5. 2026Using LLM-as-a-Judge/Jury to Advance Scalable, Clinically-Validated Safety Evaluations of Model Responses to Users Demonstrating Psychosis
  6. 2026DisaBench: A Participatory Evaluation Framework for Disability Harms in Language Models
  7. 2026PsychEthicsBench: Evaluating Large Language Models Against Australian Mental Health Ethics
  8. 2026Assessing Risks of Large Language Models in Mental Health Support: A Framework for Automated Clinical AI Red Teaming