The paper argues that accuracy is an inadequate metric for certifying a coordinator’s theory‑of‑mind in multi‑agent systems, because the required accuracy depends on deployment‑specific costs and errors are often dependent. Experiments on task assignment, Level‑Based Foraging, Hanabi, and human judgments show that accuracy can both approve harmful language‑model deployments and reject beneficial ones. The authors propose a Blind‑Spot Gate that evaluates error costs, pairwise interactions, and dependence without deployment, achieving better safety outcomes than accuracy alone.
Coordinators that act on what they infer about teammates are usually certified by the accuracy of those inferences. We argue that accuracy is the wrong certificate, for two reasons. First, the accurac...
arXiv:2606. 29654v1 Announce Type: new Abstract: Multi-agent deliberation among LLMs can improve reasoning, but deployment requires deciding when the current answer is reliable enough to act on and when it should be escalated to human review.
The paper audits a developer‑accessible on‑device language model, revealing that it can confidently produce incorrect answers while refusing benign prompts, a phenomenon termed task‑asymmetric miscalibration. The model’s confident outputs are surface‑indistinguishable, with classifiers based on user‑visible features failing to separate correct from wrong responses. The authors propose a model‑agnostic audit protocol, a surface‑indistinguishability test, and a black‑box consistency wrapper that improves reliability without requiring model access.
The study investigates how humans and AI collaborate on a puzzle task, focusing on referential uncertainty—when a description could refer to multiple objects. It finds that eliciting a belief distribution over candidate pieces yields better calibration and discrimination than raw action probabilities, and that precise descriptions or well‑targeted hedges significantly reduce the acceptance of wrong placements. However, the AI rarely externalizes uncertainty, and poorly targeted hedges can be counterproductive.
arXiv:2607. 17044v1 Announce Type: cross Abstract: Multi-step enterprise agent tasks fail in a characteristic way: single-pass inference has no checkpoint between deciding an answer and committing to it.
The paper introduces TRUST, a token‑level reward model that predicts the correctness of partial chain‑of‑thought (CoT) traces in vision‑language‑action (VLA) policies, enabling monitoring and selective steering of reasoning. On driving and manipulation VLA benchmarks, TRUST improves reasoning accuracy and reduces collision rates and trajectory errors, though its impact on overall task performance varies across tasks. The study defines two evaluation axes—correctability and actionability—to assess when CoT can serve as a runtime safety interface.
The paper treats human interventions during LLM agent failures as labels that pinpoint the internal computation that went wrong, rather than merely training data. In a controlled environment, the authors analyze one failure type—over‑asking a clarifying question when the request is already clear—using rule‑based overseer labels validated by blind human judgment. They find that the decision to ask versus act can be causally localized across several models, that over‑asking exhibits distinct mechanistic signatures, and that steering or weight consolidation can mitigate the issue, though outcomes vary by model family.
Large language model (LLM) agents now fix software, operate computers and serve customers, but their harnesses, the programs that run them, call a large LLM at every step, which makes long tasks slow...
arXiv:2606. 00005v1 Announce Type: new Abstract: We present the Consilium Protocol, a Byzantine Fault Tolerance-derived architecture for structured multi-model AI deliberation that treats inter-model disagreement as epistemic signal rather than error.
AgentAudit is an open, extensible framework that evaluates the full lifecycle of AI agents, assessing planning, tool selection, execution, memory, and reasoning across ten dimensions such as instruction integrity, security, and alignment. Unlike existing benchmarks that focus on single aspects, AgentAudit analyzes the entire execution trace to attribute failures to specific stages. The framework was tested on five large language models, revealing significant differences in trustworthiness even among models with similar task‑completion performance.
arXiv:2608. 05064v1 Announce Type: cross Abstract: Small open-weight language models increasingly run in private, offline, and cost-sensitive settings, where the key deployment question is not only what a model answers but when it should defer to a human.
The paper introduces Traverse, a benchmark of 2,518 agent trajectories and 6,967 annotated mistakes across software engineering, computer use, and science tasks, revealing that failures often go unrecovered and can cause irreversible harm before a run is deemed successful. It shows that human judges struggle to detect the first mistake in most runs, while a 4‑billion‑parameter verifier called Scout can locate failures more effectively and improve task success when used to select among candidate runs. The study demonstrates that making failure detection inexpensive and reliable can enable long‑horizon agents to learn from their own mistakes and increase trustworthiness in autonomous AI.