MagicBG logoMagicBG

AI Policy

Last updated: May 23, 2026

MagicBG is built around AI. This policy explains, in plain language, which AI systems we use, how they process your data, and the editorial rules we follow when AI assists in producing content on this site.

1. The AI in the tool

The background-removal tool on our homepage runs an open-source image-segmentation model (a quantized transformer-based foreground/background segmenter) entirely inside your browser using WebAssembly and WebGPU. Your images are processed locally on your device.

  • No image upload. Photos are never sent to our servers or any third party.
  • No training on your data. We do not collect, store, or use your images to train any model.
  • Offline-capable. After the first visit, the model can run without an internet connection.

2. Model provenance

We use openly licensed segmentation models (such as RMBG-family and SAM-2-derived models) distributed under permissive licenses. We do not fine-tune these models on user data. If a model is updated, the new weights are bundled with a site release and downloaded by your browser the next time you visit.

3. AI in editorial content

Articles on the MagicBG blog are written and reviewed by humans. We may use AI assistants (for example large language models) for limited tasks such as outlining, grammar checks, and rephrasing. Every published article is:

  • Fact-checked by a human editor against primary sources.
  • Reviewed for originality — we do not publish bulk AI-generated filler.
  • Updated when underlying tools, prices, or platform rules change.

We do not publish AI-generated photographs of real people, fabricated quotes, or fabricated statistics. Numerical claims link to a primary source whenever possible.

4. AI-generated images on the site

Some illustrative images in blog posts (diagrams, before/after mockups, decorative scenes) are produced with AI image generators. These are clearly intended as illustrations, not as documentary photographs, and are never used to misrepresent a real product, person, or location.

5. What we do not do

  • We do not run facial recognition, identification, or biometric profiling on your photos.
  • We do not sell or share images, prompts, or derived data with advertisers or data brokers.
  • We do not use generative AI to produce deepfakes of real, identifiable individuals.
  • We do not use AI to make automated decisions that produce legal or similarly significant effects about you.

6. Your rights

Because no personal data leaves your device when you use the tool, there is generally nothing for us to delete on request. For any data we do hold (for example, an email you sent us via the contact form), you can request access, correction, or deletion under our Privacy Policy.

7. Reporting issues

If you believe the tool produced a harmful, biased, or misleading output, or that an article on this site contains AI-introduced errors, please email justtome02@gmail.com with the URL and a short description. We review every report and publish corrections with a visible “Updated” date.

8. Changes to this policy

As AI capabilities and regulations evolve (EU AI Act, US state-level rules, platform policies), we will update this page. Material changes will be reflected in the “Last updated” date at the top.

Related