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

Zero Egress CAGE

Ali Emre Yenihayat, Berk Ülker · Team Zero Egress CAGE

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

Read the report

Report: Zero Egress CAGE

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

The July 2026 incident showed isolated environments fail via permitted egress. We present CAGE-1: a pre-run containment gate with 13 auditable controls. A CLI verifies evidence integrity, successfully blocking regressions in tests. CAGE-1 converts containment promises into checkable evidence.

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. I like the idea of attesting for containment policy, but I'm not sure it would have stopped cases like the HF incident where the configuration really did require exposure of the vulnerable Artifactory server.

    The text submitted reads more as a technical report than as a paper, with a lot of highly dense information (tables, lists) and not a lot of explanations, high-level descriptions, or discussions.

  2. CAGE-1 takes a smart DevSecOps approach to AI containment by shifting enforcement to a pre-run gate with 13 auditable controls. Providing a CLI tool that verifies configuration evidence and catches regressions in tests delivers immediate practical utility. However, because it operates primarily as a pre-flight check, its main blind spot is runtime state drift or in-session escalations (e.g., an agent modifying network routes post-boot). Pairing this static pre-run gate with continuous below-the-guest telemetry would make it far more comprehensive

Cite this project

@misc{yenihayat2026zero,
  title = {{Zero Egress CAGE}},
  author = {Ali Emre Yenihayat and Berk Ülker},
  year = {2026},
  month = sep,
  note = {Submitted to AI Incident Response Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/zero-egress-cage-w71t}},
  url = {https://apartresearch.com/sprints/projects/zero-egress-cage-w71t}
}

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