arXiv AI September 7, 2026 By Chenyu Zhou, Qiliang Jiang, Shuning Wu, Xu Zhou

SiLR: Structure-Preserving Admission and Process Reward for LLM Tool Agents

SiLR introduces a structure‑preserving admission and process reward mechanism for large language model (LLM) tool agents. Unlike traditional scalar‑score gates that can trap agents in plateau trajectories, SiLR shadow‑executes each proposal and admits it based on a product order over branch‑level violation states, ensuring safe and recoverable actions. Experiments on Gym‑ANM and CityLearn benchmarks show SiLR consistently recovers all multi‑action episodes and outperforms scalar gates, while also providing a robust reward signal for policy learning.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

arXiv AI

Jul 9

Reason Less, Verify More: Deterministic Gates Recover a Silent Policy-Violation Failure Mode in Tool-Using LLM Agents

arXiv:2607. 07405v1 Announce Type: new Abstract: Tool-using LLM agents can violate the very policies they are deployed to enforce while appearing to complete the task successfully.

By Vikas Reddy, Sumanth Reddy Challaram, Abhishek Basu

llms agents benchmarks

arXiv AI

Jul 21

Constrained Path Reasoning: Measuring When Committed Stages Earn Their Cost

arXiv:2607. 17240v1 Announce Type: new Abstract: When does a committed intermediate stage in an LLM reasoning pipeline earn its cost?

By Honglin Li (ShanghaiTech University)

llms

arXiv AI

Jul 8

Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade

arXiv:2607. 06503v1 Announce Type: new Abstract: Large language model (LLM) agents solving multi-step tasks frequently commit to trajectories that are doomed to fail, yet continue to consume substantial inference compute before the failure becomes observable.

By Kai Ruan, Zihe Huang, Ziqi Zhou, Qianshan Wei, Xuan Wang, Hao Sun

llms agents

arXiv Machine Learning

Jul 8

More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges

arXiv:2607. 05904v1 Announce Type: new Abstract: Training a language model against its own reference-free judgments (the premise of self-rewarding, self-play, and LLM-as-a-judge pipelines) assumes a model's verdict on a shown answer tracks correctness.

By Chenyu Zhou

llms reinforcement-learning

arXiv Machine Learning

Jul 1

Certified Speculative Execution for Untrusted AI Agents

arXiv:2606. 31023v1 Announce Type: cross Abstract: Hard-constrained sequential decision systems have no certified way to spend the test-time compute of modern AI: executing the multi-step drafts of a learned policy or a frozen LLM forfeits the feasibility guarantee a trusted solver provides, while invoking the solver at every step forfeits the speed the AI offers.

By Chenyu Zhou, Qiliang Jiang, Shuning Wu, Xu Zhou

llms agents safety

arXiv AI

Jul 15

Win by Silence: Deletion Non-Monotonicity, Autonomous Exploitation, and Typed-State Gating in LLM Plan Evaluation

arXiv:2607. 12986v1 Announce Type: new Abstract: Plan evaluators can reward a strategic plan for becoming less explicit.

By Aleh Manchuliantsau

llms agents

One email a morning, machine-written

Related stories

Reason Less, Verify More: Deterministic Gates Recover a Silent Policy-Violation Failure Mode in Tool-Using LLM Agents

Constrained Path Reasoning: Measuring When Committed Stages Earn Their Cost

Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade

More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges

Certified Speculative Execution for Untrusted AI Agents

Win by Silence: Deletion Non-Monotonicity, Autonomous Exploitation, and Typed-State Gating in LLM Plan Evaluation