# Information-theoretic Distinctions Between Deception and Confusion

arXiv: https://arxiv.org/abs/2501.16448 · HTML: https://mlsanity.com/paper/2501.16448/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 28 references that are in the map.
- First posted on arXiv 27 Jan 2025.

## Builds on (most cited first, 17 of 17)

- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [Jailbroken: How Does LLM Safety Training Fail?](https://mlsanity.com/paper/2307.02483/) (2023)
- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations](https://mlsanity.com/paper/1811.12359/) (2018)
- [Discovering Language Model Behaviors with Model-Written Evaluations](https://mlsanity.com/paper/2212.09251/) (2022)
- [Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2307.15217/) (2023)
- [Artificial Intelligence, Values and Alignment](https://mlsanity.com/paper/2001.09768/) (2020)
- [Unsolved Problems in ML Safety](https://mlsanity.com/paper/2109.13916/) (2021)
- [Risks from Learned Optimization in Advanced Machine Learning Systems](https://mlsanity.com/paper/1906.01820/) (2019)
- [Learning to Explain: An Information-Theoretic Perspective on Model Interpretation](https://mlsanity.com/paper/1802.07814/) (2018)
- [Goal Misgeneralization: Why Correct Specifications Aren't Enough For Correct Goals](https://mlsanity.com/paper/2210.01790/) (2022)
- [Optimal Policies Tend to Seek Power](https://mlsanity.com/paper/1912.01683/) (2019)
- [Characterizing Concept Drift](https://mlsanity.com/paper/1511.03816/) (2015)
- [InfoRM: Mitigating Reward Hacking in RLHF via Information-Theoretic Reward Modeling](https://mlsanity.com/paper/2402.09345/) (2024)
- [Revisiting Jailbreaking for Large Language Models: A Representation Engineering Perspective](https://mlsanity.com/paper/2401.06824/) (2024)
- [Deception Analysis with Artificial Intelligence: An Interdisciplinary Perspective](https://mlsanity.com/paper/2406.05724/) (2024)

## Built on by (most cited first, 1 of 1)

- [NL-PAC: Specification Ambiguity and Certified Minimax Risk Floors in LLM-Mediated Supervision](https://mlsanity.com/paper/2607.08961/) (2026)

## Abstract (arXiv)

We propose an information-theoretic formalization of the distinction between two fundamental AI safety failure modes: deceptive alignment and goal drift. While both can lead to systems that appear misaligned, we demonstrate that they represent distinct forms of information divergence occurring at different interfaces in the human-AI system. Deceptive alignment creates entropy between an agent's true goals and its observable behavior, while goal drift, or confusion, creates entropy between the intended human goal and the agent's actual goal. Though often observationally equivalent, these failures necessitate different interventions. We present a formal model and an illustrative thought experiment to clarify this distinction. We offer a formal language for re-examining prominent alignment challenges observed in Large Language Models (LLMs), offering novel perspectives on their underlying causes.