# Training LLMs for Honesty via Confessions

arXiv: https://arxiv.org/abs/2512.08093 · HTML: https://mlsanity.com/paper/2512.08093/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Builds on 16 of its 40 references that are in the map.
- First posted on arXiv 8 Dec 2025.

## Builds on (most cited first, 16 of 16)

- [A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions](https://mlsanity.com/paper/2311.05232/) (2023)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training](https://mlsanity.com/paper/2401.05566/) (2024)
- [Reasoning Models Don't Always Say What They Think](https://mlsanity.com/paper/2505.05410/) (2025)
- [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025)
- [Measuring short-form factuality in large language models](https://mlsanity.com/paper/2411.04368/) (2024)
- [Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety](https://mlsanity.com/paper/2507.11473/) (2025)
- [Why Language Models Hallucinate](https://mlsanity.com/paper/2509.04664/) (2025)
- [Generalizing Verifiable Instruction Following](https://mlsanity.com/paper/2507.02833/) (2025)
- [Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models](https://mlsanity.com/paper/2406.10162/) (2024)
- [LlamaFirewall: An open source guardrail system for building secure AI agents](https://mlsanity.com/paper/2505.03574/) (2025)
- [When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors](https://mlsanity.com/paper/2507.05246/) (2025)
- [Stress Testing Deliberative Alignment for Anti-Scheming Training](https://mlsanity.com/paper/2509.15541/) (2025)
- [Detecting Strategic Deception Using Linear Probes](https://mlsanity.com/paper/2502.03407/) (2025)
- [Training Language Models to Explain Their Own Computations](https://mlsanity.com/paper/2511.08579/) (2025)
- [Spilling the Beans: Teaching LLMs to Self-Report Their Hidden Objectives](https://mlsanity.com/paper/2511.06626/) (2025)

## Built on by (most cited first, 22 of 22)

- [Reward Hacking in the Era of Large Models: Mechanisms, Emergent Misalignment, Challenges](https://mlsanity.com/paper/2604.13602/) (2026)
- [A Positive Case for Faithfulness: LLM Self-Explanations Help Predict Model Behavior](https://mlsanity.com/paper/2602.02639/) (2026)
- [Predicting LLM Safety Before Release by Simulating Deployment](https://mlsanity.com/paper/2607.07184/) (2026)
- [Training Agents to Self-Report Misbehavior](https://mlsanity.com/paper/2602.22303/) (2026)
- [Hallucinations Undermine Trust; Metacognition is a Way Forward](https://mlsanity.com/paper/2605.01428/) (2026)
- [MMA: Multimodal Memory Agent](https://mlsanity.com/paper/2602.16493/) (2026)
- [Introspective Coupling: Self-Explanation Training Tracks Behavioral Change Despite Fixed Supervision](https://mlsanity.com/paper/2606.32038/) (2026)
- [Self-Attribution Bias: When AI Monitors Go Easy on Themselves](https://mlsanity.com/paper/2603.04582/) (2026)
- [Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditing](https://mlsanity.com/paper/2606.17478/) (2026)
- [Characterizing the Consistency of the Emergent Misalignment Persona](https://mlsanity.com/paper/2604.28082/) (2026)
- [Model Forensics: Investigating Whether Concerning Behavior Reflects Misalignment](https://mlsanity.com/paper/2606.26071/) (2026)
- [Stress-Testing Alignment Audits With Prompt-Level Strategic Deception](https://mlsanity.com/paper/2602.08877/) (2026)
- [Stable Reasoning, Unstable Responses: Mitigating LLM Deception via Stability Asymmetry](https://mlsanity.com/paper/2603.26846/) (2026)
- [Cognitive Fatigue in Autoregressive Transformers: Formalization and Measurement](https://mlsanity.com/paper/2605.30981/) (2026)
- [A Model of Multi-turn Human Persuadability Using Probabilistic Belief Tracing](https://mlsanity.com/paper/2606.05330/) (2026)
- [Emergent Alignment](https://mlsanity.com/paper/2606.19527/) (2026)
- [Knowing When to Quit: Diagnosing and Training LLMs to Abort Futile Reasoning](https://mlsanity.com/paper/2607.29211/) (2026)
- [When Do Models Admit They Are Wrong? Failure Disclosure Is Unstable Under Reinforcement Learning](https://mlsanity.com/paper/2609.33220/) (2026)
- [Language Models Are "Insecure" Reporters](https://mlsanity.com/paper/2609.36139/) (2026)
- [Monitorability Disposition in Large Reasoning Models](https://mlsanity.com/paper/2610.04914/) (2026)
- [Deception by Omission: Language Models Knowingly Hide Their Mistakes](https://mlsanity.com/paper/2610.11351/) (2026)
- [Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict](https://mlsanity.com/paper/2610.12360/) (2026)

## Abstract (arXiv)

Large language models (LLMs) can be dishonest when reporting on their actions and beliefs -- for example, they may overstate their confidence in factual claims or cover up evidence of covert actions. Such dishonesty may arise due to the effects of reinforcement learning (RL), where challenges with reward shaping can result in a training process that inadvertently incentivizes the model to lie or misrepresent its actions. In this work we propose a method for eliciting an honest expression of an LLM's shortcomings via a self-reported *confession*. A confession is an output, provided upon request after a model's original answer, that is meant to serve as a full account of the model's compliance with the letter and spirit of its policies and instructions. The reward assigned to a confession during training is solely based on its honesty, and does not impact positively or negatively the main answer's reward. As long as the "path of least resistance" for maximizing confession reward is to surface misbehavior rather than covering it up, this incentivizes models to be honest in their confessions. Our findings provide some justification this empirical assumption, especially in the case of egregious model misbehavior. To demonstrate the viability of our approach, we train GPT-5-Thinking to produce confessions, and we evaluate its honesty in out-of-distribution scenarios measuring hallucination, instruction following, scheming, and reward hacking. We find that when the model lies or omits shortcomings in its "main" answer, it often confesses to these behaviors honestly, and this confession honesty modestly improves with training. Confessions can enable a number of inference-time interventions including monitoring, rejection sampling, and surfacing issues to the user.