Skip to content
Apart Research
Apart Research
Contact us
Sprint projectJun 21, 2026India

Synthetic Political Speech in Regional Languages

Punith

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

Read the report

Report: Synthetic Political Speech in Regional Languages

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

The rapid advancement of AI voice cloning technology poses a novel and underexplored threat to democratic processes in linguistically diverse nations such as India. This paper proposes a comprehensive research methodology and analytical framework for investigating whether AI-generated voice clones of local political figures, including Members of Legislative Assemblies, caste association leaders, religious figures, and panchayat presidents, can manipulate trust networks in rural India more effectively than traditional text-based misinformation. We present a detailed experimental design encompassing a dataset construction protocol for 50 political leaders across 10 major Indian languages (Hindi, Bengali, Tamil, Telugu, Kannada, Malayalam, Marathi, Gujarati, Odia, and Urdu), a voice cloning pipeline using state-of-the-art text-to-speech models, and a 240-participant perception study for evaluating human detection ability and comparative persuasion impact. Grounded in existing literature on deepfake detection, rural Indian political communication, and AI safety evaluation, we derive expected outcomes and propose a contextual AI safety evaluation framework tailored to the Indian political and linguistic landscape. This paper serves as a replicable blueprint for researchers and policymakers seeking to assess the threat of synthetic political speech in multilingual democracies.

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. - Execute the Empirical Phase: Since this is currently a proposed methodology, executing the actual 240-participant perception study and constructing the physical dataset is the necessary next step to validate the hypotheses regarding the "Rural Trust Paradox".

    - Expand Beyond ElevenLabs: Relying heavily on ElevenLabs Multilingual v2 may limit the evaluation of open-source or localized models that malicious actors are more likely to self-host to evade commercial content safety guardrails.

    - Address Encrypted Platform Constraints: While the paper correctly identifies WhatsApp as the primary vector, providing specific techno-legal or media literacy intervention designs tailored to closed, encrypted networks would significantly enhance the policy actionability section.

  2. The project is explicitly proposed rather than executed, so the next step should be a small pilot study.

  3. The problem addressed is timely, important, and particularly relevant for multilingual democracies. The paper presents a thorough experimental protocol that could serve as a valuable starting point for future research, especially given its attention to regional languages and rural trust networks. However, because the submission remains primarily a proposed methodology, it is difficult to assess how the framework performs in practice. Even a small-scale pilot- such as generating a limited number of synthetic clips, evaluating audio quality, or conducting a preliminary perception study-would substantially strengthen the contribution. Additionally, discussing safeguards for responsible handling of potentially harmful synthetic political content during the research process would further reinforce the project's AI safety focus.

Cite this project

@misc{punith2026synthetic,
  title = {{Synthetic Political Speech in Regional Languages}},
  author = {Punith},
  year = {2026},
  month = jun,
  note = {Submitted to Global South AI Safety Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/synthetic-political-speech-in-regional-languages-r8op}},
  url = {https://apartresearch.com/sprints/projects/synthetic-political-speech-in-regional-languages-r8op}
}

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