Showing cs.AIShow all
2 papers · 1 filter
cs.AI2026
IatroBench: A Pre-Registered Benchmark of Clinical Omission in Language Models
David Gringras
We introduce IatroBench, a benchmark with two axes of harm (commission and omission), comprising 60 pre-registered clinical scenarios, tested on 6 models. Matched scenarios are fra…
cs.AI2026
Safety Under Scaffolding: How Evaluation Conditions Shape Measured Safety
David Gringras
Safety benchmarks usually test "bare" models that receive prompts and output responses, but real-world deployments "wrap" those models in complex scaffolds. How much do these scaff…