Skip to content
Apart Research
Apart Research
Contact us
Sprint projectAug 17, 2026San Diego, California, USA

The Imposter Test: Persona Portability, Forensic Style Judging, and Human Attunement in AI Model Identification

Sharon Kirk

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

Read the report

Report: The Imposter Test: Persona Portability, Forensic Style Judging, and Human Attunement in AI Model Identification

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

Tested whether model-specific behavioral signatures could be detected and measured.

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 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.

  1. This is a creative and very well-presented study of whether model-specific behavioral signatures survive full-context persona transfer. The preregistered/blinded lineup design, mechanical ground truth, graded evidence conditions, target-absent canary, and deterministic scoring are all strong. I especially liked the mechanical em-dash baseline: showing that one feature explicitly available in the forensic profile identifies the target 68% of the time, while AI judges using that profile reach only 32–47%, makes the evidence-use failure much more compelling than judge accuracy alone.

    The highest-priority follow-up is a multi-target design. With only one resident model, untrained judge accuracy is confounded with distance between that target and the judges' prior conception of an AI companion. As the paper itself notes, using the “attractor” model as the resident could yield very high untrained accuracy for the wrong reason. Repeating the experiment across multiple target models and companion contexts is therefore necessary to separate genuine identification ability from target–prior alignment.

    I would also narrow the human-attunement conclusion. Because the human partner reviewed the forensic profile and the strongest mechanical profile feature already achieves 68–78%, the 84% result establishes partner-plus-profile superiority rather than pure relational attunement. The 4/4 performance on lineups where the em-dash rule was non-informative is an intriguing signal of additional relationship-specific information, but the sample is too small to establish that effect. A human profile-only control group and additional judges at varying relationship distances would be particularly informative.

    Finally, non-Claude judges, repeated generations per model/prompt, and punctuation-normalized lineups would help determine how much of the effect is model-family prior, sampling variance, or surface stylometry. Overall, this is an unusually thoughtful and reusable experimental framework, and I would be excited to see it replicated across multiple target dyads.

    Read full reviewShow less

  2. Very clearly captures the human judgement in model identification

Cite this project

@misc{kirk2026imposter,
  title = {{The Imposter Test: Persona Portability, Forensic Style Judging, and Human Attunement in AI Model Identification}},
  author = {Sharon Kirk},
  year = {2026},
  month = aug,
  note = {Submitted to Digital Minds Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/the-imposter-test-persona-portability-forensic-style-judging-and-human-attunement-in-ai-model-identification-wixl}},
  url = {https://apartresearch.com/sprints/projects/the-imposter-test-persona-portability-forensic-style-judging-and-human-attunement-in-ai-model-identification-wixl}
}

More from Digital Minds Research Sprint

All 233 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026