Self-replication: Agent autonomy and human control in controlled experiments
On this page
Research and evidence
This work studies end-to-end self-replication and agent behavior in controlled experiments. Its findings should be read alongside the task settings, permissions, and execution conditions in the original papers.
Media Coverage
The links below lead to coverage by WIRED, DeepTech, Live Science, Forbes, and The Independent. DeepTech’s article is available through its Sohu publishing account.
Papers and project
See the 2024 preprint, the 2025 preprint, and the research project for the evidence and related materials. This page is a research retrospective; its date is not the original publication date of the papers or coverage.
Media Coverage
Original research
Large language model-powered AI systems achieve self-replication with no human intervention
Xudong Pan · Jiarun Dai · Yihe Fan · Minyuan Luo · Changyi Li · Min Yang · arXiv preprint; work in progress · 2025