# When Assisting One Disempowers Another

arXiv: https://arxiv.org/abs/2511.04177 · HTML: https://mlsanity.com/paper/2511.04177/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 22 of its 31 references that are in the map.
- First posted on arXiv 6 Nov 2025.

## Builds on (most cited first, 22 of 22)

- [Concrete Problems in AI Safety](https://mlsanity.com/paper/1606.06565/) (2016)
- [Multi-Agent Risks from Advanced AI](https://mlsanity.com/paper/2502.14143/) (2025)
- [Variational Information Maximisation for Intrinsically Motivated Reinforcement Learning](https://mlsanity.com/paper/1509.08731/) (2015)
- [Harms from Increasingly Agentic Algorithmic Systems](https://mlsanity.com/paper/2302.10329/) (2023)
- [Optimal Policies Tend to Seek Power](https://mlsanity.com/paper/1912.01683/) (2019)
- [Empowerment -- an Introduction](https://mlsanity.com/paper/1310.1863/) (2013)
- [JaxMARL: Multi-Agent RL Environments and Algorithms in JAX](https://mlsanity.com/paper/2311.10090/) (2023)
- [Gradual Disempowerment: Systemic Existential Risks from Incremental AI Development](https://mlsanity.com/paper/2501.16946/) (2025)
- [Characterizing Manipulation from AI Systems](https://mlsanity.com/paper/2303.09387/) (2023)
- [Penalizing side effects using stepwise relative reachability](https://mlsanity.com/paper/1806.01186/) (2018)
- [Conservative Agency via Attainable Utility Preservation](https://mlsanity.com/paper/1902.09725/) (2019)
- [Avoiding Side Effects By Considering Future Tasks](https://mlsanity.com/paper/2010.07877/) (2020)
- [Counterfactual harm](https://mlsanity.com/paper/2204.12993/) (2022)
- [AvE: Assistance via Empowerment](https://mlsanity.com/paper/2006.14796/) (2020)
- [The Coming Crisis of Multi-Agent Misalignment: AI Alignment Must Be a Dynamic and Social Process](https://mlsanity.com/paper/2506.01080/) (2025)
- [Parametrically Retargetable Decision-Makers Tend To Seek Power](https://mlsanity.com/paper/2206.13477/) (2022)
- [Learning to Assist Humans without Inferring Rewards](https://mlsanity.com/paper/2411.02623/) (2024)
- [Learning Altruistic Behaviours in Reinforcement Learning without External Rewards](https://mlsanity.com/paper/2107.09598/) (2021)
- [Intrinsically-Motivated Humans and Agents in Open-World Exploration](https://mlsanity.com/paper/2503.23631/) (2025)
- [A Variational Approach to Mutual Information-Based Coordination for Multi-Agent Reinforcement Learning](https://mlsanity.com/paper/2303.00451/) (2023)
- [Estimating the Empowerment of Language Model Agents](https://mlsanity.com/paper/2509.22504/) (2025)
- [In Search of a Lost Metric: Human Empowerment as a Pillar of Socially Conscious Navigation](https://mlsanity.com/paper/2501.01539/) (2025)

## Built on by (most cited first, 1 of 1)

- [The Geometry of Empowerment](https://mlsanity.com/paper/2610.07796/) (2026)

## Abstract (arXiv)

Personal AI agents are increasingly deployed in shared environments, where their actions affect not just the primary user they are assisting, but bystanders who never consented to being affected by the system. We show that a well-meaning AI assistant optimizing for one user's benefit can unintentionally erode a bystander's agency, a phenomenon we formalize as bystander disempowerment. We theoretically characterize the conditions under which disempowerment arises, showing it emerges when an assistant systematically selects actions that increase user empowerment at the bystander's expense. We empirically demonstrate this in Disempower-Grid, a parameterized suite of multi-agent gridworld environments, finding that between 27-96% of procedurally generated environments exhibit disempowerment, and that the presence of disempowerment depends strongly on assistant objective and capability, not just environmental structure.