# The Wrong Kind of Right: Quantifying and Localizing Misfired Alignment in LLMs

arXiv: https://arxiv.org/abs/2606.18656 · HTML: https://mlsanity.com/paper/2606.18656/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 34 of its 50 references that are in the map.
- First posted on arXiv 17 Jun 2026.

## Builds on (most cited first, 30 of 34)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Proximal Policy Optimization Algorithms](https://mlsanity.com/paper/1707.06347/) (2017)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Chain-of-Thought Prompting Elicits Reasoning in Large Language Models](https://mlsanity.com/paper/2201.11903/) (2022)
- [Direct Preference Optimization: Your Language Model is Secretly a Reward Model](https://mlsanity.com/paper/2305.18290/) (2023)
- [Deep reinforcement learning from human preferences](https://mlsanity.com/paper/1706.03741/) (2017)
- [Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2204.05862/) (2022)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Learning to summarize from human feedback](https://mlsanity.com/paper/2009.01325/) (2020)
- [Transformer Feed-Forward Layers Are Key-Value Memories](https://mlsanity.com/paper/2012.14913/) (2020)
- [Interpretability in the Wild: a Circuit for Indirect Object Identification in GPT-2 small](https://mlsanity.com/paper/2211.00593/) (2022)
- [A General Language Assistant as a Laboratory for Alignment](https://mlsanity.com/paper/2112.00861/) (2021)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [BBQ: A Hand-Built Bias Benchmark for Question Answering](https://mlsanity.com/paper/2110.08193/) (2021)
- [Bias and Fairness in Large Language Models: A Survey](https://mlsanity.com/paper/2309.00770/) (2023)
- [XSTest: A Test Suite for Identifying Exaggerated Safety Behaviours in Large Language Models](https://mlsanity.com/paper/2308.01263/) (2023)
- [Safety Alignment Should Be Made More Than Just a Few Tokens Deep](https://mlsanity.com/paper/2406.05946/) (2024)
- [Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification](https://mlsanity.com/paper/1903.04561/) (2019)
- [Bias in Bios: A Case Study of Semantic Representation Bias in a High-Stakes Setting](https://mlsanity.com/paper/1901.09451/) (2019)
- [AI Alignment: A Comprehensive Survey](https://mlsanity.com/paper/2310.19852/) (2023)
- [OR-Bench: An Over-Refusal Benchmark for Large Language Models](https://mlsanity.com/paper/2405.20947/) (2024)
- [SORRY-Bench: Systematically Evaluating Large Language Model Safety Refusal](https://mlsanity.com/paper/2406.14598/) (2024)
- [Gender bias and stereotypes in Large Language Models](https://mlsanity.com/paper/2308.14921/) (2023)
- [A Mechanistic Understanding of Alignment Algorithms: A Case Study on DPO and Toxicity](https://mlsanity.com/paper/2401.01967/) (2024)
- [Mitigating the Alignment Tax of RLHF](https://mlsanity.com/paper/2309.06256/) (2023)
- [A Meta-Transfer Objective for Learning to Disentangle Causal Mechanisms](https://mlsanity.com/paper/1901.10912/) (2019)
- [Interventional Causal Representation Learning](https://mlsanity.com/paper/2209.11924/) (2022)
- [Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable](https://mlsanity.com/paper/2503.00555/) (2025)
- [A Survey on Fairness in Large Language Models](https://mlsanity.com/paper/2308.10149/) (2023)
- [WinoQueer: A Community-in-the-Loop Benchmark for Anti-LGBTQ+ Bias in Large Language Models](https://mlsanity.com/paper/2306.15087/) (2023)

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

- [When Protection Overrides Consequences: Understanding Protection Mode in Large Language Model Decision-Making](https://mlsanity.com/paper/2610.19948/) (2026)

## Abstract (arXiv)

Warning: This paper studies stereotypes and biases, and contains potentially disturbing examples, used for illustration purposes only. Our findings should not be interpreted as an argument against alignment. Instead, this paper highlights the need for principled approaches to more advanced alignment. Alignment aims to ensure that large language models (LLMs) behave safely and reliably, including by avoiding unsafe inferences. However, we show that such safety-oriented behaviors can misfire: models may reject warranted conclusions even when they are explicitly supported by context. We call this failure mode misfired alignment, where alignment-induced changes cause LLMs to override explicit evidence. To quantify this phenomenon, specifically on stereotype-related alignment, we introduce VETO, a benchmark consisting of 2,032 BBQ-derived contrastive pairs, and define a new metric, Misfired Alignment Rate (MAR), which measures on a 0 to 100 scale how often a model fails on a stereotype-related question but succeeds on its contrastive counterpart. We benchmark 25 LLMs on VETO, and show that all LLMs, including the most recent ones, exhibit non-trivial (4.7 to 18.9%) MARs while all human participants achieve 0.0% MAR. Controlled priming experiments further show that alignment-induced cues can substantially amplify MAR across LLMs, indicating that these failures are not merely artifacts of individual examples but can be induced by safety-related framing. Mechanistic analyses on open-weight LLMs reveal late-layer suppression of evidence-supported answers, and comparisons between instruct and base LLMs suggest that this suppression emerges after instruction training. These findings show that current alignment methods can overgeneralize surface-level safety cues, to the point of overriding objective evidence, motivating more work on alignment objectives that better preserve contextual grounding.