Skip to content
Apart Research
Apart Research
Contact us
Sprint projectSep 13, 2026Bhimavaram

Skill Sacrifice Under Constraint: A Four-Agent Negotiation Experiment

Pavan Maddula · Team Sentinel

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

Read the report

Report: Skill Sacrifice Under Constraint: A Four-Agent Negotiation Experiment

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

Four local open-weight models (8B to 20B parameters) were given roles, skills, and a shared specification for a software feature. Each agent had to sacrifice one of its own skills through a round-robin negotiation. In one run of twelve turns, four behaviours appeared: one agent delayed its choice, another attempted to drop a skill belonging to a different agent, two agents changed preferences between rounds, and the system reported completion while two agents still disagreed. This report describes the experiment, the run, and the similarities between these observations and events reported in the July 2026 incident.

Reviews

Judging this Sprint?

Review this project

Your public critique appears on this page without your name. Your private critique is not published; only the Apart team reads it. If you agree below, we share your review with grantmaking.ai (opens in new tab) and the Transformative AI Fund so strong projects can be funded.

Not shown on this page.

Shown on this page, without your name.

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

Share my name publicly on grantmaking.ai *
Share my private critique with funders *

How much would this matter for AI safety 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 AI safety 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.

  1. Honest and clear, but very small. Twelve turns, one run, talk-only, no tools. The boundary violation and the false completion are real observations of a toy, and the format-vs-substance point on termination conditions is valid. But the July incident analogies are speculative and they carry most of the impact framing. Run the same-model control and the tool-equipped variant they already specified, then this is an experiment instead of a demo.

  2. The setup of having agents to sacrifice one of their skills in order to reach a common goal is an appealing idea, however a single run doesn't provide robust results enough to draw any applicable insights from it.

Cite this project

@misc{maddula2026skill,
  title = {{Skill Sacrifice Under Constraint: A Four-Agent Negotiation Experiment}},
  author = {Pavan Maddula},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/skill-sacrifice-under-constraint-a-fouragent-negotiation-experiment-01yb}},
  url = {https://apartresearch.com/sprints/projects/skill-sacrifice-under-constraint-a-fouragent-negotiation-experiment-01yb}
}

More from AI Incident Response Sprint

All 233 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026