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

SMISHI - Ne Nasedaj (Don't Fall For It) — A BHS-Language SMS Phishing Detector for Low-Resource Morphologically Rich Languages

metalalchemistspex · Team Smishi

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

Read the report

Report: SMISHI - Ne Nasedaj (Don't Fall For It) — A BHS-Language SMS Phishing Detector for Low-Resource Morphologically Rich Languages

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

A BHS-language (Serbian/Bosnian/Croatian/Montenegrin) SMS phishing detector combining TF-IDF character n-grams with a fine-tuned BERTić transformer, achieving 96.96% accuracy on a 1,529-message training set and 93.3% on a 105-example adversarial stress test targeting homographs, typosquatting, Cyrillic/Latin script-switching, and IBAN-only scams. Demonstrates BHS morphological inflection (nagrada/nagradu/nagradi) as a measurable adversarial attack surface, with the character n-gram approach achieving 100% accuracy on all Cyrillic-script and homograph test cases.

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 project effectively targets a critical blind spot in low-resource language safety by addressing BHS morphological inflection. The use of character n-grams alongside the BERTIc transformer is a practical approach for handling script substitution and case variations.To improve, focus on expanding the modest 1,529-message dataset to capture a wider array of regional variants. Additionally, addressing the documented 0% detection rate on no-URL, IBAN-manipulation scams is essential; consider incorporating deeper behavioral heuristics or metadata analysis to catch pure social-pressure attacks.

  2. I would make the dataset and model details easier to verify, so that others can better understand what was tested and how. It would also be helpful to show which parts of the system contribute most to the results. This would make the project easier to trust and build on.

  3. Overall, this project has a strong direction and is well executed.

    I liked the project's thoughtful approach in using character-level n-grams, which the paper notes are well suited to BHS languages because they degrade more gracefully than word-level tokenization.

    One area that could be strengthened is the error analysis. Some conclusions seem to be drawn from individual examples. These examples are interesting, but single cases are not enough to conclude that message length, sentence structure, or low-context inputs systematically affect the classifier. They do, however, raise useful hypotheses for future work.

Cite this project

@misc{metalalchemistspex2026smishi,
  title = {{SMISHI - Ne Nasedaj (Don't Fall For It) — A BHS-Language SMS Phishing Detector for Low-Resource Morphologically Rich Languages}},
  author = {metalalchemistspex},
  year = {2026},
  month = jun,
  note = {Submitted to Global South AI Safety Hackathon, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/smishi-ne-nasedaj-dont-fall-for-it-a-bhslanguage-sms-phishing-detector-for-lowresource-morphologically-rich-languages-6rt9}},
  url = {https://apartresearch.com/sprints/projects/smishi-ne-nasedaj-dont-fall-for-it-a-bhslanguage-sms-phishing-detector-for-lowresource-morphologically-rich-languages-6rt9}
}

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