Text analysisCybersecurity

ScamLLM

A classifier from an IEEE S&P study: it separates ordinary requests to a language model from requests that order text and a page meant to deceive people. A human makes the decision on a disputed operation; automatic blocking without review is not acceptable.

Developer
University of Texas at Arlington, USA
First release
Dec 2023
Latest release
Aug 2025
Sizes
about 125M (based on RoBERTa-base)
License
Commercial use with conditionsthe model card states unknown — the license is undefined
Russian
Not supported
Running
On your own serverAlso runs without a GPU
Industries
Security, Finance
Deploy on your server Hugging Face

What it does

  • Flagging assistant prompts that look like preparation for deception
  • Checking campaign texts before they are sent
  • Filling internal filtering rules with examples

Where it is used

Information securityBanks and payment servicesPlatforms with user-generated content

Hardware requirements

LaptopLaptop or regular PC, up to 8 GB of VRAM — smaller versions
fits
1 GPUOne GPU with 16–80 GB — mid-size versions
no versions
ClusterServer with several GPUs — flagship versions
no versions

Versions

  1. Обновление репозитория
  2. ScamLLM

How to run it

On your own serverWeights are downloaded from Hugging Face and served with vLLM (load and API) or llama.cpp (modest hardware). The model then runs in a closed network with no per-request fees.Hugging Face
How much hardware you needCalculate the VRAM for the model size, context length and number of concurrent requests.Open the hardware calculator

I can set this up end to end: pick the model size, deploy it on your server and connect it to your systems.

Frequently asked questions

Can ScamLLM be used in a commercial project?

With conditions. License: the model card states unknown — the license is undefined. Restrictions vary — region, company revenue, attribution requirements. Have a lawyer check the terms before a commercial launch.

What hardware does ScamLLM need?

At minimum: Laptop or regular PC, up to 8 GB of VRAM — smaller versions. Some versions also run on an ordinary CPU, without a GPU. You can calculate the exact VRAM for your model size and context in the hardware calculator.

Does ScamLLM support Russian?

No. The developer documentation lists the supported languages and Russian is not among them.

Where can I download ScamLLM and what does it cost?

The ScamLLM weights are open and free to download. You only pay for the hardware it runs on and for the setup. Source links are at the bottom of this page.

How I deploy it for clients

  1. SelectionI pick the model size for your task and hardware and test it on your examples.
  2. DeploymentI deploy it on your server or in a closed network and provide an API.
  3. Fine-tuningI fine-tune it on your data (LoRA) or connect a knowledge base — whichever is cheaper for the task.
  4. IntegrationI connect it to your CRM, ERP, bot, website or team chat and set up monitoring.

Similar models

CybersecurityPhishing Email Detection DistilBERTcybersectony · not disclosedCommercial use allowed

A very light classifier for emails and links showing signs of phishing. It errs in both directions, so borderline emails are still reviewed by a person.

DetailsText analysisFlowX ScamGuardFlowX.AI · USA and RomaniaCommercial use allowed

Small models from a banking software company that read the text of an SMS, email or chat message and explain which scam tactic it uses. A human makes the decision on a disputed operation; automatic blocking without review is not acceptable.

DetailsModeration and safetyAprielGuardServiceNow · USACommercial use allowed

A guard model that catches both harmful content and attacks on AI (prompt injection, jailbreaks), including when agents use tools.

Details

Source: huggingface.co/phishbot/ScamLLM. Data checked against the model card on 22 Sep 2026. Have a lawyer review the license before commercial launch.