Skip to content
Apart Research
Apart Research
Contact us
Sprint projectMay 24, 2026Stockholm

Verified But Wrong

Philip Nilsson

Submitted to The Secure Program Synthesis Hackathon. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Verified But Wrong

Open in a new tab (opens in new tab)
Code (opens in new tab)More on verified-but-wrong-demo.vercel.app (opens in new tab)
Share

Verified But Wrong studies a target-validity failure mode in vericoding: formally verified code can satisfy a supplied specification while violating the natural-language intent that specification was meant to capture. We audit a public Dafny vericoding benchmark, validate 11 externally sourced demonstrations including 3 direct-Dafny cases, and show in a controlled benchmark that incomplete specs can select wrong implementations. The core recommendation is simple: before using a formal spec as a candidate-selection target, vericoding pipelines should audit whether the spec is actually the right target.

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. Aligning natural language with formal specifications is a challenging but important task. This work introduces an additional step to identify discrepancies between the natural language instructions and the generated specifications. However, the methodology for detecting these gaps and ensuring the reliability of this check is insufficiently explained. The verification process appears to rely heavily on model-based evaluation; incorporating a simple, deterministic symbolic check would significantly enhance the system's reliability.

Cite this project

@misc{nilsson2026verified,
  title = {{Verified But Wrong}},
  author = {Philip Nilsson},
  year = {2026},
  month = may,
  note = {Submitted to The Secure Program Synthesis Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/verified-but-wrong-izy3}},
  url = {https://apartresearch.com/sprints/projects/verified-but-wrong-izy3}
}

More from The Secure Program Synthesis Hackathon

All 61 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026