# Safety of self-evolving systems

Area: [Synthetic data and self-improvement for LLMs](https://mlsanity.com/synth/index.md). HTML: https://mlsanity.com/synth/safety-of-self-evolving-systems/

## Key facts

- Safety of self-evolving systems: 47 papers on arXiv since 2023 (data through 14 Oct 2026).
- 4 of its 47 mapped papers (9%) cite “Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections” (2026, arXiv 2602.15654).
- 3 of its 47 mapped papers (6%) cite “Chasing Moving Targets with Online Self-Play Reinforcement Learning for Safer Language Models” (2025, arXiv 2506.07468).
- 3 new papers 17 Sep–14 Oct.

## Most cited here

- [Zombie Agents: Persistent Control of Self-Evolving LLM Agents via Self-Reinforcing Injections](https://mlsanity.com/paper/2602.15654/) (2026), cited by 4 papers here
- [Chasing Moving Targets with Online Self-Play Reinforcement Learning for Safer Language Models](https://mlsanity.com/paper/2506.07468/) (2025), cited by 3 papers here
- [SEAS: Self-Evolving Adversarial Safety Optimization for Large Language Models](https://mlsanity.com/paper/2408.02632/) (2024), cited by 2 papers here
- [OEP: Poisoning Self-Evolving LLM Agents via Locally Correct but Non-Transferable Experiences](https://mlsanity.com/paper/2605.18930/) (2026), cited by 2 papers here
- [Efficient Toxic Content Detection by Bootstrapping and Distilling Large Language Models](https://mlsanity.com/paper/2312.08303/) (2023), cited by 1 papers here
- [Trojan's Whisper: Stealthy Manipulation of OpenClaw through Injected Bootstrapped Guidance](https://mlsanity.com/paper/2603.19974/) (2026), cited by 1 papers here
- [Membrane: A Self-Evolving Contrastive Safety Memory for LLM Agent Defense](https://mlsanity.com/paper/2606.05743/) (2026), cited by 1 papers here
- [Safety in Self-Evolving LLM Agent Systems: Threats, Amplification, and Case Studies](https://mlsanity.com/paper/2606.23075/) (2026), cited by 1 papers here

## New (17 Sep–14 Oct)

- [Hidden in Plain Sight: Advancing Harmful Content Understanding via Perception-Aware Adversarial Self-Play](https://mlsanity.com/paper/2610.15449/) (2026-10-11)
- [SkillPoison: Progressive Skill Poisoning via Successful Experiences](https://mlsanity.com/paper/2610.07645/) (2026-10-06)
- [SelfOp: An Optimization Algorithm for Self-Improving Security Agents](https://mlsanity.com/paper/2609.22792/) (2026-09-19)