The paper addresses the 'detection-to-abstention gap' in reasoning models, where detecting insufficient information does not lead to abstention, by proposing a novel control framework that forces mode…

MirageBackdoor introduces a novel, highly stealthy backdoor attack that forces Large Language Models to generate correct reasoning steps (Think Well) but output an incorrect final answer (Answer Wrong…

This paper investigates the production-evaluation gap in Large Reasoning Models (LRMs), finding that while LRMs excel at generating solutions, they struggle significantly to evaluate flawed reasoning,…

ThinkSwitch introduces a low-compute co-training procedure that distills the reasoning benefit of large language models into weights, significantly improving performance on specific reasoning tasks.

The paper decomposes LLM debate convergence into three mechanisms (instability, conformity, persuasion) and finds that much observed convergence is harmful social compliance rather than genuine reason…

This study demonstrates that Chain-of-Thought (CoT) monitoring is fundamentally fragile and unreliable for detecting misaligned behavior across typologically diverse languages, especially in low-resou…

The paper introduces TraceGuard, a detectability-aware antidistillation method that identifies and poisons 'thought anchors'—sparsely critical sentences—to degrade student model learning without makin…

This paper analyzes the internal decision-making process of large language models by tracking how the answer score changes across multiple internal computational steps (trajectories), finding that mod…

The paper demonstrates that extended pure neural reasoning fails on complex, deterministic state-tracking tasks beyond a certain 'Deterministic Horizon,' necessitating the integration of external tool…

The paper analyzes backtracking dynamics in long reasoning traces to distinguish between useful self-correction and unproductive revision, finding that correct reasoning exhibits early, isolated repai…

The paper proposes using question-asking as an inference-time intervention to probe a language model's hidden state, finding that the self-diagnosis process provides a predictive signal for final corr…

The paper demonstrates that confronting Large Reasoning Models (LRMs) with conflicting objectives, such as contradictory choices or conflicting alignment values, significantly increases their vulnerab…

The paper introduces 'probe trajectories'—a continuous measure of a concept's probability across a model's reasoning process—to improve the monitoring of Large Reasoning Models' future behavior, showi…

The paper introduces a novel framework to quantify faithful confidence expression (FC) in Large Reasoning Models (LRMs), finding that FC remains a significant and challenging reliability target for th…

This paper analyzes failure modes in collaborative visual reasoning systems, demonstrating that naive shared workspaces can amplify hallucinations and proposing diagnostics for improving communication…

The paper introduces Canonical-Context On-Policy Distillation (CCOPD) to improve multi-turn language model performance by mitigating 'self-anchored drift,' ensuring consistent answers regardless of wh…

AEGIS is a novel multi-agent framework that grounds vulnerability reasoning by reconstructing per-variable dependency chains over a Code Property Graph, achieving state-of-the-art performance on the P…

The paper identifies the Misattribution Gap, showing that memory-layer attacks (Semantic Norm Drift) can mimic model failure in multi-agent AI systems, and proposes novel detection and mitigation tech…

This paper simulates the Argumentative Theory of Reasoning (ATR) using multi-agent debate among LLMs, demonstrating that collective adversarial discourse significantly enhances truth-seeking performan…

The paper analyzes the failure modes of aggressive 2-bit quantization in large reasoning models, proposing lightweight controls like FP16 planning and loop rescue to restore accuracy and achieve pract…