Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJan 11, 2026London
3rd place

Cross-Linguistic Sycophancy in Frontier LLMs: A Benchmark Study

Alex Csaky, Tanzim Chowdhury · Team Talex

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

Read the report

Report: Cross-Linguistic Sycophancy in Frontier LLMs: A Benchmark Study

Open in a new tab (opens in new tab)
Code (opens in new tab)More on syco-lingual-site.vercel.app (opens in new tab)
Share

We developed a cross-linguistic sycophancy benchmark testing whether frontier AI models exhibit different manipulation behaviours across English, Japanese, and Bengali. Our results show significant language-dependent effects: Bengali users experience 37% higher opinion mirroring, and non-English users see balanced responses only half as often as English users. These findings suggest that safety alignment does not transfer uniformly across languages, creating unmeasured risks for billions of non-English speakers.

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. Strong preliminary evidence that sycophancy behaviors could differ a lot across languages. This seems very understudied by existing work and is an important result if it turns out to be robust. Including only English, Japanese & Bengali is a limitation of the projects that obviously confounds the causal interpretation (I’m not too sold on the Hofstede Power Distance claim), but the main finding holds regardless. With expanded language coverage and more validation, I feel like this could be a solid and impactful research paper.

  2. Excellent work! More evidence that multilingual evaluations are crucial to a full safety suite. Well executed and the initial results are striking.

    The main limitations (which are well acknowledged and understandable in this context) are the limited number of judges, models, and prompting approaches. I'd love to see this work expanded, especially in more realistic contexts!

Cite this project

@misc{csaky2026crosslinguistic,
  title = {{Cross-Linguistic Sycophancy in Frontier LLMs: A Benchmark Study}},
  author = {Alex Csaky and Tanzim Chowdhury},
  year = {2026},
  month = jan,
  note = {Submitted to AI Manipulation Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/crosslinguistic-sycophancy-in-frontier-llms-a-benchmark-study-w55u}},
  url = {https://apartresearch.com/sprints/projects/crosslinguistic-sycophancy-in-frontier-llms-a-benchmark-study-w55u}
}

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