The paper reports that counterfactual fairness audits of clinical language‑model agents are unreliable without accounting for a per‑action instability floor. By repeatedly running identical vignettes, the authors found that actions changed 8.7% of the time, with instability varying eightfold across actions. A second model confirmed a pooled floor of 6.7%, showing that any reported fairness estimate lacking this floor cannot be interpreted as evidence of disparity.
By Rohith Reddy Bellibaltu, Manpreet Singh, Deepak Parashar, Rahul Joshi
arXiv:2606. 16723v1 Announce Type: new Abstract: Large language model (LLM) agents increasingly take actions (screening applicants, recommending credit, triaging patients), yet fairness for LLMs is still measured by grading answers.
By Triveni Morla, Rohith Reddy Bellibaltu, Manpreet Singh, Manmeet Singh Kapoor
arXiv:2608. 06949v1 Announce Type: new Abstract: Prior benchmarking work has shown that a single large language model (LLM), forced to make life-or-death resource-allocation decisions, exhibits measurable demographic bias.
By Paul-Peter Arslan
The paper investigates how language‑model judges can make version‑dependent errors when evaluating upgraded agents. Using 35 public coding‑agent submissions, two customer‑service agents, and over a thousand expert‑labeled trajectories, the authors show that fixed judges often reject task‑conditioned error invariance and can incorrectly approve failed patches, especially as agent capability increases. Paired audits of current outputs reduce interval width only marginally, and the study concludes that independent human patch review is still necessary.
By Jiapeng Li
The paper introduces a new evaluation method called "same-input rerun" to assess the consistency of clinical language‑model agents across repeated runs. By replaying 1,000 MedAgentBench tasks with identical inputs, the authors find that action‑level outputs—such as test orders, medication requests, and referrals—vary significantly, even when benchmark scores remain unchanged. The study demonstrates that current benchmarks, which typically evaluate only a single run per task, can miss substantial behavioral divergence.
By Rohith Reddy Bellibatlu, Manpreet Singh, Zhoutian Han, Wenbin Zhang
arXiv:2607. 18828v1 Announce Type: new Abstract: Readiness stress-testing of medical AI has focused on closed-ended and multimodal benchmarks.
By Koyar Afrasyab
arXiv:2606. 10315v1 Announce Type: cross Abstract: LLM-as-judge is the default instrument for evaluating conversational agents, yet its reliability is almost always reported as agreement with human ratings, not recall of real defects.
By Sawyer Zhang, Alexander Wang, Sophie Lei
arXiv:2608. 03744v1 Announce Type: new Abstract: Clinical decision support is moving toward committees of language-model agents deliberating on a shared workspace.
By Sebasti\'an Andr\'es Cajas Ord\'o\~nez, Agastya Munnangi, Aldo Marzullo, Felipe Ocampo Osorio, Quang Bui, Mohammad Shahin, Armaan Grewal, Emmanuel Paul Kwesiga, Anqi Peter Li, Josephine Nanyonjo, Aaditya Panchal, Arshnoor Bhutani, Nikhil Jaiswal, Milit S. Patel, Maximin Lange, Leo Anthony Celi
arXiv:2607. 28608v1 Announce Type: new Abstract: Clinical risk models routinely achieve strong aggregate performance while producing materially different error rates across patient subgroups.
By Sparsh Roy, Samuel Girmachew, Nishita Chavan
The paper investigates whether automatic safety judges evaluate the content of a model’s reply or merely its style. By keeping the reply content fixed and adding various style wrappers—such as educational disclaimers, fake reasoning blocks, or token refusals—the authors show that many judges flip their verdicts, indicating that style can influence safety judgments. The study evaluates over 600 jailbreak examples across multiple judges, revealing that some judges are highly susceptible to style-based manipulation while others remain robust.
By Yongxi Zhou, Wenbo Ye, Yuanzhe Liu, Zihan Dong, Junwei Yao
The paper investigates whether automatic safety judges evaluate the content of a model’s reply or merely its style. By adding content‑invariant style wrappers—such as educational disclaims or token refusals—to fixed replies, the authors show that many judges flip their verdicts, revealing exploitable blind spots. Across more than 600 jailbreak examples and eight judges, some judges exhibit high flip rates (e.g., GPT‑4o‑mini 19.9%) while others remain largely stable, and human validation confirms that most flips are judge errors rather than content changes.
arXiv:2608.23397v1 Announce Type: new
Abstract: Interactive clinical agents must gather decisive evidence and convert it into grounded actions under partial observability. A correct final diagnosis a...
By Ruoyu Wu, Shenfu Xie, Yinqian Sun, Haibo Tong, Feifei Zhao