# Self-Authored Verification Is Unreliable in Heuristic Self-Improving Agents

arXiv: https://arxiv.org/abs/2607.24300 · HTML: https://mlsanity.com/paper/2607.24300/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Self-evolving LLM agents, one of 13 directions in Synthetic data and self-improvement for LLMs.
- Builds on 19 of its 23 references that are in the map.
- First posted on arXiv 27 Jul 2026.

## Placed in

- Synthetic data and self-improvement for LLMs / [Self-evolving LLM agents](https://mlsanity.com/synth/self-evolving-llm-agents/)

## Builds on (most cited first, 19 of 19)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [SWE-bench: Can Language Models Resolve Real-World GitHub Issues?](https://mlsanity.com/paper/2310.06770/) (2023)
- [DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence](https://mlsanity.com/paper/2606.19348/) (2026)
- [Large Language Models Cannot Self-Correct Reasoning Yet](https://mlsanity.com/paper/2310.01798/) (2023)
- [AlphaEvolve: A coding agent for scientific and algorithmic discovery](https://mlsanity.com/paper/2506.13131/) (2025)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [Revisiting the Arcade Learning Environment: Evaluation Protocols and Open Problems for General Agents](https://mlsanity.com/paper/1709.06009/) (2017)
- [When Can LLMs Actually Correct Their Own Mistakes? A Critical Survey of Self-Correction of LLMs](https://mlsanity.com/paper/2406.01297/) (2024)
- [BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games](https://mlsanity.com/paper/2411.13543/) (2024)
- [CoEvoSkills: Self-Evolving Agent Skills via Co-Evolutionary Verification](https://mlsanity.com/paper/2604.01687/) (2026)
- [The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence](https://mlsanity.com/paper/2605.26494/) (2026)
- [Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity](https://mlsanity.com/paper/2607.00248/) (2026)
- [Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use](https://mlsanity.com/paper/2605.02964/) (2026)
- [Revisit Self-Debugging with Self-Generated Tests for Code Generation](https://mlsanity.com/paper/2501.12793/) (2025)
- [ReVeal: Self-Evolving Code Agents via Reliable Self-Verification](https://mlsanity.com/paper/2506.11442/) (2025)
- [Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR](https://mlsanity.com/paper/2603.07084/) (2026)
- [Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement](https://mlsanity.com/paper/2507.18742/) (2025)
- [When AI Reviews Its Own Code: Recursive Self-Training Collapse in Code LLMs](https://mlsanity.com/paper/2606.28438/) (2026)

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

- [ERRAND: Budgeted Maintenance of Agent Memory](https://mlsanity.com/paper/2609.29545/) (2026)
- [When Better Gets Worse: Improvement Fidelity for Self-Improving Agents in Adaptive Worlds](https://mlsanity.com/paper/2609.32677/) (2026)
- [Which Self-Improvements Should We Trust? Reliable Self-Improvement When Agents Reuse Their Benchmarks](https://mlsanity.com/paper/2609.33180/) (2026)
- [Reliable Memory from Unstable Growth: Curating Self-Written Agent Playbooks with Randomized Experiments](https://mlsanity.com/paper/2610.18427/) (2026)

## Abstract (arXiv)

Self-improving agents accumulate capability by repeatedly rewriting procedural policies, controllers, or heuristic rules. They typically rely on self-authored tests or metrics to decide whether to accept subsequent edits. The agent controls both the optimized object and its verifier. As a result, self-assigned scores can remain near perfect while real deployment performance degrades or stays low. We study this problem through the verifier--deployment gap. This gap refers to the discrepancy between an agent's self-authored verification signal and a sealed deployment evaluation that the agent cannot observe or access. We ask how self-authored verification fails under iterative policy-and-test rewriting, how the failure changes with capability, and how little exogenous trust is sufficient to prevent real regressions from being deployed. To address this problem, we introduce a Sealed Exogenous Acceptance Loop (SEAL). SEAL retains self-authored tests but compares each candidate with the incumbent through a fixed harness-side audit. The agent cannot author or inspect the audit, receives only accept/reject, and the whole incumbent state is retained after a clear regression. Our experiments show that this problem often appears in heuristic learning settings. These settings require trial-and-error discovery of the target objective. We further find that failures of self-written verification are stratified by capability. Weaker agents tend to damage previously acquired strategies behind easy self-tests. Stronger agents are more stable, but they still mismeasure the deployment distribution. Standard self-written constraints do not reliably close this gap. In contrast, SEAL outperforms unprotected baselines across six models and three random seeds. Reliable self-improvement need not abandon self-verification, but it requires at least one deployment-acceptance signal outside the agent's control.