The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models
Masked diffusion language models (MDMs) uniquely support any-order generation, with confidence-based decoding currently serving as the de facto standard inference policy. To optimize for this, recent training schemes attempt to align training mask patterns directly with those observed during generation. However, we argue that confidence-based decoding is inherently misaligned with the logical-flow trajectories required for complex reasoning, and that confidence-aligned training actively entrench
Record details
Published: 27 May 2026
Source: arXiv
Category: Research
Topics: Regulation · Safety & alignment
Retrieved: 14 July 2026
Related evidence
These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.
Return-to-Go Is More Than a Number: Q-Guided Alignment for Return-Conditioned Supervised Learning
arXiv · 27 May 2026
Relevance as a Vulnerability: How Web Retrieval Degrades Safety Alignment in LLM Agents
arXiv · 28 May 2026
DREAM-R: Multimodal Speculative Reasoning with RL-Based Refined Drafting, Precise Verification, and Fully Parallel Execution
arXiv · 27 May 2026
SafeMed-R1: Clinician-Audited Safety and Ethics Alignment for Medical Large Language Models
arXiv · 27 May 2026
Beyond Attack Success Rate: Temporal Logit Observability for LLM Safety Failures
arXiv · 28 May 2026
AI Loss of Control Incident Management: Response & Resilience
arXiv · 28 May 2026
How to cite this record
ethics.ai (27 May 2026), “The Confidence Shortcut: A Reasoning Failure Mode of Masked Diffusion Models,” evidence record 3557, https://ethics.ai/record/3557 (originally published by arXiv).
Use and limitations
This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.