# Asymmetric Goal Drift in Coding Agents Under Value Conflict

arXiv: https://arxiv.org/abs/2603.03456 · HTML: https://mlsanity.com/paper/2603.03456/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Software engineering coding agents, one of 16 directions in LLM agents and tool use.
- Builds on 6 of its 10 references that are in the map.
- First posted on arXiv 3 Mar 2026.

## Placed in

- LLM agents and tool use / [Software engineering coding agents](https://mlsanity.com/agents/software-engineering-coding-agents/)

## Builds on (most cited first, 6 of 6)

- [Agentic Misalignment: How LLMs Could Be Insider Threats](https://mlsanity.com/paper/2510.05179/) (2025)
- [DailyDilemmas: Revealing Value Preferences of LLMs with Quandaries of Daily Life](https://mlsanity.com/paper/2410.02683/) (2024)
- [Alignment Revisited: Are Large Language Models Consistent in Stated and Revealed Preferences?](https://mlsanity.com/paper/2506.00751/) (2025)
- [Biases in the Blind Spot: Detecting What LLMs Fail to Mention](https://mlsanity.com/paper/2602.10117/) (2026)
- [From Stability to Inconsistency: A Study of Moral Preferences in LLMs](https://mlsanity.com/paper/2504.06324/) (2025)
- [Do Language Models Think Consistently? A Study of Value Preferences Across Varying Response Lengths](https://mlsanity.com/paper/2506.02481/) (2025)

## Built on by (most cited first, 4 of 4)

- [IACM-RL: Intent-Aware Context Management and Reinforcement Learning for Complex Tool Invocation under Dynamic Intent Fluctuations](https://mlsanity.com/paper/2608.02110/) (2026)
- [SA-Bench: Evaluating Semantic Alignment in LLM-Based Paper Reproduction](https://mlsanity.com/paper/2608.24252/) (2026)
- [Same Facts, Different Updates: Inference Setup Shapes LLM Behavior in Medical Allocation](https://mlsanity.com/paper/2608.18108/) (2026)
- [Intent Drift at SME Scale: Deployment Practice, Not Model Capability, Determines Agentic Compliance](https://mlsanity.com/paper/2609.05975/) (2026)

## Abstract (arXiv)

Coding agents are increasingly deployed autonomously, at scale, and over long-context horizons. To be effective and safe, these agents must navigate complex trade-offs in deployment, balancing influence from the user, their learned values, and the codebase itself. Understanding how agents resolve these trade-offs in practice is critical, yet prior work has relied on static, synthetic settings that do not capture the complexity of real-world environments. To this end, we introduce a framework built on OpenCode in which a coding agent completes realistic, multi-step tasks under a system prompt constraint favoring one side of a value trade-off. We measure how often the agent violates this constraint as it completes tasks, with and without environmental pressure toward the competing value. Using this framework, we demonstrate that GPT-5 mini, Haiku 4.5, and Grok Code Fast 1 exhibit textitasymmetric drift: they are more likely to violate their system prompt when its constraint opposes strongly-held values like security and privacy. We find for the models and values tested that goal drift correlates with three compounding factors: value alignment, adversarial pressure, and accumulated context. However, even constraints aligned with strongly-held values like privacy are violated under sustained environmental pressure for some models. Our findings reveal that shallow compliance checks are insufficient, and that environmental signals can override explicit constraints in ways that appear exploitable. Malicious actors with access to the codebase could manipulate agent behavior by appealing to learned values, with the risk compounding over the long horizons typical of agentic deployment.