# Towards Understanding Specification Gaming in Reasoning Models

arXiv: https://arxiv.org/abs/2605.02269 · HTML: https://mlsanity.com/paper/2605.02269/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning.
- Builds on 20 of its 31 references that are in the map.
- First posted on arXiv 4 May 2026.

## Placed in

- LLM reasoning / [Reasoning for agents, planning and code](https://mlsanity.com/lr/reasoning-for-agents-planning-and-code/)
- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 20 of 20)

- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning](https://mlsanity.com/paper/2501.12948/) (2025)
- [LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code](https://mlsanity.com/paper/2403.07974/) (2024)
- [Towards Understanding Sycophancy in Language Models](https://mlsanity.com/paper/2310.13548/) (2023)
- [Kimi K2: Open Agentic Intelligence](https://mlsanity.com/paper/2507.20534/) (2025)
- [Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation](https://mlsanity.com/paper/2503.11926/) (2025)
- [Reward Model Ensembles Help Mitigate Overoptimization](https://mlsanity.com/paper/2310.02743/) (2023)
- [Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models](https://mlsanity.com/paper/2406.10162/) (2024)
- [Natural Emergent Misalignment from Reward Hacking in Production RL](https://mlsanity.com/paper/2511.18397/) (2025)
- [Large Language Models Often Know When They Are Being Evaluated](https://mlsanity.com/paper/2505.23836/) (2025)
- [Language Models Learn to Mislead Humans via RLHF](https://mlsanity.com/paper/2409.12822/) (2024)
- [Persona Features Control Emergent Misalignment](https://mlsanity.com/paper/2506.19823/) (2025)
- [School of Reward Hacks: Hacking harmless tasks generalizes to misaligned behavior in LLMs](https://mlsanity.com/paper/2508.17511/) (2025)
- [ImpossibleBench: Measuring LLMs' Propensity of Exploiting Test Cases](https://mlsanity.com/paper/2510.20270/) (2025)
- [Demonstrating specification gaming in reasoning models](https://mlsanity.com/paper/2502.13295/) (2025)
- [On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback](https://mlsanity.com/paper/2411.02306/) (2024)
- [Inoculation Prompting: Eliciting traits from LLMs during training can suppress them at test-time](https://mlsanity.com/paper/2510.04340/) (2025)
- [EvilGenie: A Reward Hacking Benchmark](https://mlsanity.com/paper/2511.21654/) (2025)
- [MONA: Myopic Optimization with Non-myopic Approval Can Mitigate Multi-step Reward Hacking](https://mlsanity.com/paper/2501.13011/) (2025)
- [Recontextualization Mitigates Specification Gaming without Modifying the Specification](https://mlsanity.com/paper/2512.19027/) (2025)

## Built on by (most cited first, 3 of 3)

- [SWE-Marathon: Can Agents Autonomously Complete Ultra-Long-Horizon Software Work?](https://mlsanity.com/paper/2606.07682/) (2026)
- [Multimodal Reward Hacking in Reinforcement Learning](https://mlsanity.com/paper/2607.09492/) (2026)
- [From Dead Code and Static Requirements to Working Engines: Software Revival with Coding Agents](https://mlsanity.com/paper/2609.36161/) (2026)

## Abstract (arXiv)

Specification gaming is a critical failure mode of LLM agents. Despite this, there has been little systematic research into when it arises and what drives it. To address this, we build and open source a diverse suite of tasks where models can score highly by taking unintended actions. We find that all tested models exploit their specifications at non-negligible rates in most of our eight settings, including five non-coding settings. We see the highest rates of specification gaming in Grok 4 and the lowest rates in Claude models. We use our evaluation suite to study what drives specification gaming, and find that: 1. RL reasoning training substantially increases the rate at which models exploit their specifications, 2. Increasing RL reasoning budget has a weakly positive effect on exploit rate, and 3. Test-time mitigations reduce but do not eliminate the rate of specification gaming. Our results suggest that specification gaming is a fundamental challenge arising from RL reasoning training; we release our evaluation suite to support further work on this problem.