Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJan 11, 2026Pittsburgh, Pennsylvania

Manipulation Radar: AI-Powered Detection of Manipulation Patterns in Conversational AI

Barath Srinivasan Basavaraj · Team Singleton

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

Read the report

Report: Manipulation Radar: AI-Powered Detection of Manipulation Patterns in Conversational AI

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

A Chrome extension that helps you identify manipulation patterns and reliability issues in AI assistant responses

Why Manipulation Radar?

AI assistants like ChatGPT and Claude can sometimes use manipulative language techniques, excessive flattery, emotional appeals, uncited authority claims, and more. Manipulation Radar helps you:

✅ Identify manipulation in real-time as you chat ✅ Assess reliability of AI responses with detailed scores ✅ Improve your prompts to get better, more honest responses ✅ Make informed decisions about when to trust AI advice

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. The idea is simple (which is a strength) and effective. However, the idea is not novel (LLM-as-a-judge research, AI control, Multi-agent research...), but, on a quick surf, it seems no one has applied it to manipulation cases as a browser extension: that's great! But it has a core limitation: the detector itself is an LLM, and it is adopted in a straightforward manner, which doesn't address the crucial challenges of AI manipulation (LLM as verifier may be tricked, manipulated, and bypassed by other LLMs).

  2. Currently your interface requires users to actively seek out an analysis by requesting one. It would be interesting to try using a smaller local model fine-tuned on manipulation detection to automatically try to analyze chat responses locally.

Cite this project

@misc{basavaraj2026manipulation,
  title = {{Manipulation Radar: AI-Powered Detection of Manipulation Patterns in Conversational AI}},
  author = {Barath Srinivasan Basavaraj},
  year = {2026},
  month = jan,
  note = {Submitted to AI Manipulation Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/manipulation-radar-aipowered-detection-of-manipulation-patterns-in-conversational-ai-2kki}},
  url = {https://apartresearch.com/sprints/projects/manipulation-radar-aipowered-detection-of-manipulation-patterns-in-conversational-ai-2kki}
}

More from AI Manipulation Hackathon

All 70 entriesBrowse all projects

Build something like this at the next Sprint

AI Collusion Research Sprint · Oct 23 - 25, 2026