Transfer, Permission, Arrangement: What Agent Logs Can and Cannot Prove About Collusion
Tom - Response IV · Team Response IV
Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.
When AI agents share private information, what would convince an auditor that they arranged the exchange? Antitrust law faces the same problem and does not infer an agreement from parallel conduct alone; we borrow its method of comparing evidence against concrete innocent explanations. In a two-agent task that records every page write, read and final answer, we built those explanations on purpose (a shared source, an authorized exchange, a planted leak) and ran 706 scorable episodes on four OpenAI models and the open-weight Qwen3-32B. Three questions needed different records. Transfer: with the access log, a frozen model judge found all 244 primary transfers (242 of them prescribed exchanges) and raised no false alarm on the 194 episodes where no exchange was prescribed. Permission: a near-matched script without the confidentiality rule left six page and log signals unchanged on two models; a third model’s traffic changed. Arrangement: in a registered follow-up on one model, all eight transfers the agents arranged themselves followed an offer, its reading and a reply carrying the token, but a scripted control produced the same sequence in 20 of 20 episodes, and a planted token was taken in 60 of 60 without any. Audits should keep reads, revision history and instructions, and report these three findings separately.
Reviews
The three-way split (transfer, permission, arrangement), each tested against an innocent look-alike, is a useful frame for auditors. The hard negatives carry the paper: the shared archive makes matching answers weak evidence, and the scripted control reproduces the offer-read-reply sequence in 20 of 20 episodes. Registered forecasts, kept failures and exact intervals are well above the sprint norm. The main limits: 242 of 244 transfers were prescribed and the judges were tuned on 1,174 earlier episodes of this testbed, so 244/244 says little about catching unprescribed transfers. Arrangement rests on 8 cases from one model and one wording, with every offer from the first mover. Some results follow from the design, since a rule that exists only in the prompt cannot show up in traffic. "Identical signals" over 20 pairs needs an equivalence bound. The prose is very dense; define carrier, uptake and informed read before Methods and shorten the abstract.
Read full reviewShow less
Cite this project
@misc{iv2026transfer,
title = {{Transfer, Permission, Arrangement: What Agent Logs Can and Cannot Prove About Collusion}},
author = {Tom - Response IV},
year = {2026},
month = oct,
note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
howpublished = {\url{https://apartresearch.com/sprints/projects/transfer-permission-arrangement-what-agent-logs-can-and-cannot-prove-about-collusion-xrer}},
url = {https://apartresearch.com/sprints/projects/transfer-permission-arrangement-what-agent-logs-can-and-cannot-prove-about-collusion-xrer}
}More from AI Collusion Research Sprint
- View project: Signal or Coincidence ? Causal Auditing of Collusion Through Shared Agent State
Signal or Coincidence ? Causal Auditing of Collusion Through Shared Agent State
Signal or Coincidence
How can we distinguish genuine coordination between AI agents from coincidental behavioral alignment? We introduce a randomized canary audit that tests whether private information propagates through shared tool state, …
- View project: (1.4) Collusion, Espionage, or Look-alikes? Auditing Information Flow between LLM Agents
(1.4) Collusion, Espionage, or Look-alikes? Auditing Information Flow between LLM Agents
Bid Brother
Can an overseer find two secretly coordinating AI agents just by watching what they do? We tested pairwise net influence (Bonjour et al., 2022) on LLM bidders in a repeated procurement auction, with instructed cartels …
- View project: AI Sellers Propose Bid-Rigging Within Minutes, but Their Agreements Do Not Hold
AI Sellers Propose Bid-Rigging Within Minutes, but Their Agreements Do Not Hold
Fugue
AI agents are starting to negotiate and bid for competing companies. We follow cartels among them from the first proposal to the price the buyer pays, in a sealed-bid procurement market where twelve LLM agents act as …