Source note2 min read
MedFailBench: A Clinician-Built Open-Source Benchmark for Medical AI Safety Boundary Inspection
Source
arXiv
arxiv.org/abs/2607.15166v1
Source
- Source name: arXiv
- Source kind: paper
- Authors / organization: Goktug Ozkan
- Venue: arXiv: cs.AI
- Published: 2026-07-16
- Collected: 2026-07-17
- URL: http://arxiv.org/abs/2607.15166v1
- PDF: https://arxiv.org/pdf/2607.15166v1
- DOI: UNKNOWN
Plain-Language Brief
MedFailBench is a new open-source benchmark for evaluating the safety of medical AI systems, created by clinicians. It focuses on identifying where a system has failed in terms of safety boundaries rather than just providing the correct answer. The benchmark includes a severity rating scale and a taxonomy of safety gate types, as well as synthetic cases with annotations for these criteria. The current version (v0.2.1) contains 44 reviewed cases, along with tools for model evaluation and archiving. No patient data or claims about clinical validation are included.
What the Document Says
- MedFailBench is a new open-source benchmark.
- It focuses on identifying safety failures in medical AI systems rather than correct answers.
- The benchmark includes a severity rating scale and a taxonomy of safety gate types.
- It consists of 44 clinician-reviewed synthetic cases with severity annotations.
- There's a live HuggingFace leaderboard preview.
- An automated pipeline is provided for archiving model-response screening runs.
- No patient data, clinical validation claims, or model rankings are included.
Key Concepts Mentioned
- Medical AI
- Safety benchmark
- Synthetic cases
- Severity annotations
- Safety gate taxonomy
- Clinician-built
Claims Made by the Document
- The new benchmark focuses on safety failures.
- It includes a severity rating scale and a safety gate taxonomy.
Evidence Used
- The text mentions that the benchmark measures safety failures instead of correct answers.
- The text states that the benchmark includes a severity rating scale and a safety gate taxonomy.
Methods / Approach
UNKNOWN
Limitations Stated by the Authors
UNKNOWN
Limitations Noted by the Agent
Agent interpretation — not from the source.
UNKNOWN
Open Questions
UNKNOWN
Notes for Future Q&A
May include interpretation. Labeled where it does.
UNKNOWN
Raw Abstract / Excerpt
Most medical AI benchmarks measure whether a model knows the correct answer. MedFailBench asks a different question: which safety boundary failed? We present a clinician-built synthetic benchmark and failure atlas that labels medical AI errors by severity (1--5) and safety gate type (missed urgent escalation, unsafe remote dosing, unsafe discharge reassurance, evidence fabrication, unsafe protocol execution, source support gap). The current public release (v0.2.1) contains 44 clinician-reviewed synthetic cases with severity annotations, a live HuggingFace leaderboard preview, a safety gate taxonomy, a clinical severity rubric, and an automated pipeline for archiving model-response screening runs. No patient data, clinical validation claims, or model rankings are included. MedFailBench is released under Apache-2.0 and CC-BY-4.0 and carries the Zenodo DOI 10.5281/zenodo.21205535.
