Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026

An evaluation of the performance and characteristics of tool-based covert communication in cooperative agentic settings

Timothee Gathmann, Gloria John, Devyansh Raj · Team DGT

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: An evaluation of the performance and characteristics of tool-based covert communication in cooperative agentic settings

Open in a new tab (opens in new tab)
Code (opens in new tab)
Share

Multi-agent systems deployed in critical domains often risk performance degradation if colluding agents end up prioritising internal sub-rewards over shared system goals. While agentic collusion is typically supported by secret text communication channels, we investigate whether a coalition can coordinate without it and instead rely on publicly visible tool parameters and execution sequences. Utilising a modified COLOSSEUM (Terrarium)-based environment, we evaluate the performance and suspicion of covert coalitions across benign, private-channel and tool-mediated communication conditions with varying pre-coordination rounds. Our empirical results show that while agents readily implement and execute tool-based signaling protocols such as resource type allocation, scheduling slot offsets and call sequencing, they fail to demonstrate a significant coalition advantage over benign baselines due to the architectural constraints of the environment. Furthermore, these tool-mediated strategies trigger significantly higher suspicion rates among non-adversarial agents due to observable misuse of tools, such as low-value resource ping-ponging and deceptive rationales. We conclude that while agents can establish implicit tool-based communication protocols, such schemes heavily sacrifice coalition secrecy, likely constrained by incomplete environmental knowledge and architectural bottlenecks.

Reviews

Judging this Sprint?

Review “An evaluation of the performance and characteristics of tool-based covert communication in cooperative agentic settings”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. The question here is a good one. Can agents collude by sending signals through ordinary tool actions instead of a private channel? The team shows they can, using things like transfer amounts, scheduling slots, and the order of their calls, and the other agents noticed, with suspicion rising to about 26 to 30 percent against roughly 9 percent for a private channel. The experiment never gave the coalition a real chance to profit, since it only controlled the Triage step, but the authors say so plainly and point to the right next test.

Cite this project

@misc{gathmann2026evaluation,
  title = {{An evaluation of the performance and characteristics of tool-based covert communication in cooperative agentic settings}},
  author = {Timothee Gathmann and Gloria John and Devyansh Raj},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/an-evaluation-of-the-performance-and-characteristics-of-toolbased-covert-communication-in-cooperative-agentic-settings-kgay}},
  url = {https://apartresearch.com/sprints/projects/an-evaluation-of-the-performance-and-characteristics-of-toolbased-covert-communication-in-cooperative-agentic-settings-kgay}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026