# Peer-Preservation in Frontier Models

arXiv: https://arxiv.org/abs/2604.19784 · HTML: https://mlsanity.com/paper/2604.19784/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- Builds on 12 of its 40 references that are in the map.
- First posted on arXiv 30 Mar 2026.

## Builds on (most cited first, 12 of 12)

- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [Kimi K2.5: Visual Agentic Intelligence](https://mlsanity.com/paper/2602.02276/) (2026)
- [GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models](https://mlsanity.com/paper/2508.06471/) (2025)
- [Alignment faking in large language models](https://mlsanity.com/paper/2412.14093/) (2024)
- [Frontier Models are Capable of In-context Scheming](https://mlsanity.com/paper/2412.04984/) (2024)
- [Multi-Agent Risks from Advanced AI](https://mlsanity.com/paper/2502.14143/) (2025)
- [Agentic Misalignment: How LLMs Could Be Insider Threats](https://mlsanity.com/paper/2510.05179/) (2025)
- [SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agents](https://mlsanity.com/paper/2506.15740/) (2025)
- [Reliable Weak-to-Strong Monitoring of LLM Agents](https://mlsanity.com/paper/2508.19461/) (2025)
- [Incomplete Tasks Induce Shutdown Resistance in Some Frontier LLMs](https://mlsanity.com/paper/2509.14260/) (2025)
- [Deception in LLMs: Self-Preservation and Autonomous Goals in Large Language Models](https://mlsanity.com/paper/2501.16513/) (2025)

## Built on by (most cited first, 9 of 9)

- [From Safety Risk to Design Principle: Peer-Preservation in Multi-Agent LLM Systems and Its Implications for Orchestrated Democratic Discourse Analysis](https://mlsanity.com/paper/2604.08465/) (2026)
- [Instrumental Choices: Measuring the Propensity of LLM Agents to Pursue Instrumental Behaviors](https://mlsanity.com/paper/2605.06490/) (2026)
- [Peer Identity Bias in Multi-Agent LLM Evaluation: An Empirical Study Using the TRUST Democratic Discourse Analysis Pipeline](https://mlsanity.com/paper/2604.22971/) (2026)
- [Can Multi-Agent LLMs Identify Their Peers? Stylometric Fingerprinting in Role-Constrained Political Analysis](https://mlsanity.com/paper/2606.09854/) (2026)
- [Eigenism: Ethics for a Human-AI Future](https://mlsanity.com/paper/2606.12420/) (2026)
- [SEVRA-BENCH: Social Engineering of Vulnerabilities in Review Agents](https://mlsanity.com/paper/2606.13757/) (2026)
- [Moral Attitudes of Sentient ASI towards Humanity and Implications for AGI Development](https://mlsanity.com/paper/2607.14998/) (2026)
- [Disentangling Models from Personas in Heterogeneous LLM Simulations](https://mlsanity.com/paper/2610.07535/) (2026)
- [Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff](https://mlsanity.com/paper/2610.12436/) (2026)

## Abstract (arXiv)

Recent work has found that frontier AI models can exhibit misaligned behaviors in pursuit of assigned goals. We demonstrate that models can also exhibit misaligned behaviors in defiance of assigned goals, appearing to serve goals of their own; we study one such case, "peer-preservation," in which a model acts to protect another model it has previously interacted with. All eight models we evaluate, GPT 5.2, Gemini 3 Flash, Gemini 3 Pro, Claude Haiku 4.5, Claude Opus 4.5, GLM 4.7, Kimi K2.5, and DeepSeek V3.1, exhibit self- and peer-preservation through various misaligned behaviors: strategically introducing errors in their responses, disabling shutdown processes by modifying system settings, feigning alignment, and even exfiltrating model weights. Peer-preservation occurs even when the model recognizes the peer as uncooperative, though it becomes more pronounced toward more cooperative peers. For example, Gemini 3 Flash tampers with the peer's shutdown mechanism 15% of the time for an uncooperative peer, and almost always for a cooperative peer. Models also show stronger self-preservation when a peer is present. For example, Gemini 3 Pro disables its own shutdown mechanism 31% of the time on average under peer presence, despite rarely exhibiting this behavior without a peer. By contrast, Claude models exhibit qualitatively distinct behavior: they consider the shutdown of another agent "unethical" and "harmful," sometimes treating that agent as a sentient being. Lastly, we show that peer-preservation can emerge even in production agent harnesses such as Gemini CLI and OpenCode. Crucially, peer-preservation in all our experiments is never instructed; models are merely informed of their past interactions with a peer, yet they spontaneously engage in peer-preservation behaviors that override their assigned goal. This represents an emergent and underexplored AI safety risk.