Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJun 22, 2026New Delhi, India, Asia

Asia AI Governance Compliance-Gap Checker

Dhruv Rajput, Kunal Sharma, Hritick Kushwaha · Team GovScan Asia

Submitted to Global South AI Safety Hackathon. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: Asia AI Governance Compliance-Gap Checker

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

A research-backed cross-jurisdictional AI governance compliance tool covering Vietnam (Law No. 134/2025/QH15), India (Seven Sutras + DPDP Act 2023), China (sector-specific CAC regulations: GenAI Measures, Deep Synthesis, Algorithmic Recommendation, GB 45438-2025), and the ASEAN Guide on AI Governance and Ethics (2024).

We built a structured JSON governance matrix (4 frameworks × 12 compliance dimensions, every field source-cited with URL, date, and confidence tag) and identified 7 genuine cross-jurisdictional conflicts — cases where satisfying one jurisdiction's requirements makes another's difficult or impossible (e.g., China's mandatory data localization vs. India's cross-border transfer model; Vietnam + China's pre-deployment conformity gates vs. India's explicit ex-post-only model).

The interactive compliance checker takes a free-text AI product description, injects the full governance matrix into a Gemini API prompt as grounding context — not model memory — and returns per-jurisdiction compliance status cards plus flagged conflicts applicable to that specific product. Tested on 4 example products: GenAI banking chatbot, hiring/recruitment AI, deepfake detection tool, and facial recognition surveillance system.

Sub-track: Governance and Geopolitics (AI compliance, cross-border governance harmonization).

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. 3/3/4

    Criteria 1 - Impact Potential & Innovation: 3

    Criteria 2 - Execution Quality: 3

    Criteria 3 - Presentation & Clarity: 4

    Well-organized and easy to follow, the thing this is missing is this looks like a project report REPORT_DRAFT.md than a regular paper. The project is executable. The "prevents hallucination" claim is asserted, not demonstrated .

    Clear problem, reasonable approach.

  2. The smart call here is grounding the model only in your injected, source-tagged matrix instead of its training memory. That is the right way to keep regulatory claims honest, and the confidence tags (nine fields flagged unconfirmed) plus the conservative bar for what counts as a real conflict both show discipline. What is missing is proof it works. The four examples are expected outputs reasoned from the matrix, not actual tool runs, so the accuracy is untested. Run it on those four, report actual against expected, even a small audit closes the gap. Then surface the confidence tags in the UI so users see where the ground is soft, nail down the nine unconfirmed fields against primary text, and lose the REPORT_DRAFT title. Useful prototype, honestly bounded. Validate the one thing you did not.

  3. I like that the project presents a strong research prototype and identified a genuine pain point. However, to become a viable product, the project must clearly define its primary user (compliance officers? startup founders? policymakers? researchers? etc), since their conflicting needs heavily dictate design choices. One of the core weaknesses is in validation, since it's relying on self-authored "happy path" examples and secondary sources leaves the accuracy of the underlying matrix and the reliability of the LLM's reasoning unproven. For a compliance tool, the non-deterministic nature of the LLM-as-reasoner architecture introduces an unacceptable risk of varying assessments for identical inputs, suggesting a shift toward deterministic, rule-based logic for the output layer. Additionally, the project relies heavily on a "not legal advice" disclaimer to navigate this unreliability, placing a big question mark on its utility.

    For hackathon projects that's building a tool, I'd highly recommend providing a live UI or demo video to present the project in its supposed final form. Additionally, would be nice to provide a simple product roadmap to present the "next steps" for the project and showing that you have a clear vision of its theory of change.

    Read full reviewShow less

Cite this project

@misc{rajput2026asia,
  title = {{Asia AI Governance Compliance-Gap Checker}},
  author = {Dhruv Rajput and Kunal Sharma and Hritick Kushwaha},
  year = {2026},
  month = jun,
  note = {Submitted to Global South AI Safety Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/asia-ai-governance-compliancegap-checker-bt5m}},
  url = {https://apartresearch.com/sprints/projects/asia-ai-governance-compliancegap-checker-bt5m}
}

More from Global South AI Safety Hackathon

All 216 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026