mlsanityData to 14 Oct 2026

Taxonomy of Comprehensive Safety for Clinical Agents

arXiv 2509.22041 ↗First posted 26 Sep 2025cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Safety is a paramount concern in clinical chatbot applications, where inaccurate or harmful responses can lead to serious consequences. Existing methods--such as guardrails and tool calling--often fall short in addressing the nuanced demands of the clinical domain. In this paper, we introduce TACOS (TAxonomy of COmprehensive Safety for Clinical Agents), a fine-grained, 21-class taxonomy that integrates safety filtering and tool selection into a single user intent classification step. TACOS is a taxonomy that can cover a wide spectrum of clinical and non-clinical queries, explicitly modeling varying safety thresholds and external tool dependencies. To validate our taxonomy, we curate a TACOS-annotated dataset and perform extensive experiments. Our results demonstrate the value of a new taxonomy specialized for clinical agent settings, and reveal useful insights about train data distribution and pretrained knowledge of base models.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 16 of its 41 references that are in the map. First posted on arXiv 26 Sep 2025.

Builds on

16 of 41 references.

References · 16

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  3. 2022Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedbackcited by 3,720
  4. 2023Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversationscited by 1,207
  5. 2023Prompt Injection attack against LLM-integrated Applicationscited by 605
  6. 2023NeMo Guardrails: A Toolkit for Controllable and Safe LLM Applications with Programmable Railscited by 365
  7. 2024Introducing v0.5 of the AI Safety Benchmark from MLCommonscited by 62
  8. 2024Conformal Prediction for Natural Language Processing: A Surveycited by 53
  9. 2020KR-BERT: A Small-Scale Korean-Specific Language Modelcited by 17
  10. 2024The Need for Guardrails with Large Language Models in Medical Safety-Critical Settings: An Artificial Intelligence Application in the Pharmacovigilance Ecosystemcited by 11
  11. 2023Intent Identification and Entity Extraction for Healthcare Queries in Indic Languagescited by 9
  12. 2024Enhancing Guardrails for Safe and Secure Healthcare AIcited by 7
4 more
  1. 2021Self-Diagnosis through AI-enabled Chatbot-based Symptom Checkers: User Experiences and Design Considerationscited by 4
  2. 2019Transfer Learning Robustness in Multi-Class Categorization by Fine-Tuning Pre-Trained Contextualized Language Modelscited by 1
  3. 2024The Right Model for the Job: An Evaluation of Legal Multi-Label Classification Baselinescited by 1
  4. 2024Specialty detection in the context of telemedicine in a highly imbalanced multi-class distributioncited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025MATRIX: Multi-Agent simulaTion fRamework for safe Interactions and conteXtual clinical conversational evaluation
  2. 2024Ensuring Safety and Trust: Analyzing the Risks of Large Language Models in Medicine
  3. 2025Large language models provide unsafe answers to patient-posed medical questions
  4. 2026CareGuardAI: Context-Aware Multi-Agent Guardrails for Clinical Safety & Hallucination Mitigation in Patient-Facing LLMs
  5. 2025CARES: Comprehensive Evaluation of Safety and Adversarial Robustness in Medical LLMs
  6. 2026When Medical Safety Alignment Fails: A Benchmark for Evaluating LLMs on High-Risk Medical Queries
  7. 2025Balancing Safety and Helpfulness in Healthcare AI Assistants through Iterative Preference Alignment
  8. 2026SafePro: Evaluating the Safety of Professional-Level AI Agents