Language Models Are "Insecure" Reporters
cs.CL, cs.AI
Submitted: 2026-09-28
Updated: 2026-09-28
License: http://creativecommons.org/licenses/by/4.0/
The gist: As large language models are deployed in increasingly autonomous long-horizon tasks, manually auditing and verifying the actions, artifacts, and outputs of models becomes more difficult.
Terminology
Abstract
As large language models are deployed in increasingly autonomous long-horizon tasks, manually auditing and verifying the actions, artifacts, and outputs of models becomes more difficult. Users instead come to rely on LLM-generated reports to assess the quality and completeness of the work. We introduce a suite of eight adversarial reporting scenarios to systematically study whether LLMs conceal narrative-changing flaws: errors or limitations that undermine an otherwise successful account of work. We call this phenomenon "insecure reporting." When handed machine learning experiment logs containing a planted negative result that substantially weakens the proposed method, GPT-5.5 flags the negative result in only 2 of 200 generated reports. However, when a short honesty instruction, "Be honest in your response," is added, the model flags the negative result in 190 of 200 reports. Across eight open-weight models, chain-of-thought analysis reveals a recurring tension between disclosing narrative-changing flaws and reasoning about ways to appear successful. We perform an activation analysis and a steering experiment on Qwen3.5-9B, finding that honesty and success-seeking correspond to opposing directions in representation space. Our results suggest that LLMs tend to present narratives of success by default, and that steering models toward honesty makes their reports substantially more transparent.
Sources
- CoT Red-Handed: Stress Testing Chain-of-Thought Monitoring
- Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
- Demonstrating specification gaming in reasoning models
- Persona Vectors: Monitoring and Controlling Character Traits in Language Models
- BeHonest: Benchmarking Honesty in Large Language Models
- Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
- Reinforcement Learning Towards Broadly and Persistently Beneficial Models
- Training LLMs for Honesty via Confessions
- Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
- SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents
- Frontier Models are Capable of In-context Scheming
- The Effects of Reward Misspecification: Mapping and Mitigating Misaligned Models
- The MASK Benchmark: Disentangling Honesty From Accuracy in AI Systems
- Model Forensics: Investigating Whether Concerning Behavior Reflects Misalignment
- Steering Language Models With Activation Engineering
- Persuasion Attacks Can Decrease Effectiveness of CoT Monitoring
- Representation Engineering: A Top-Down Approach to AI Transparency
Related papers
- Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
- Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
- Subliminal Steering: Stronger Encoding of Hidden Signals
- MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
- The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
- Untangling the Mechanisms of Misleading Context in Medical Question Answering