Takedown Copilot

AI tool that turns online impersonation into action, generates platform reports, DMCA notices, and step by step takedown letters in under 2 minutes. No lawyer needed.

Comment

Inspiration

Every week, creators and everyday people discover AI generated fakes of themselves online, deepfake videos, voice clones, fake accounts. The scary part isn't finding it. It's not knowing what to do next. Most people freeze. They don't know what a DMCA notice is, where to report it, or how to write a professional takedown request. So the fake content stays up.

I built Takedown Copilot to fix that exact moment of panic.

What it does

Takedown Copilot is a free, no login tool that turns online impersonation into action in under 2 minutes. You tell it your name, the platform, the type of impersonation (deepfake video, voice clone, fake profile, or AI image), and paste the link. It generates three things instantly:

  • A short blurb ready to paste into the platform's in app report form
  • A full formal takedown email with subject line, legal language, and your details
  • A step by step checklist of what to do right now, in order

How we built it

  • Frontend: Pure HTML, CSS, and JavaScript
  • AI: Gemini API (with a smart offline fallback template so it always works)
  • Hosting: Firebase Hosting, deployed and live
  • Search: Google Custom Search API for finding public mentions of impersonation

The whole app is one file. That was intentional, fast to load, easy to audit, works on any device.

Challenges we ran into

Getting the AI to return clean, structured JSON every time was tricky. We solved it with a strict prompt format and a fallback template that kicks in silently if the API call fails, so the user always gets a result.

We also had to think carefully about tone. The output needed to sound professional and calm, not robotic, because the person using it is probably stressed and scared.

Accomplishments that we're proud of

  • Built and deployed a fully working app in under 48 hours with zero backend servers
  • The offline fallback means the app never breaks, users always get a result even if the AI call fails
  • Every output sounds human and professional, not like it came from a template
  • One single HTML file that does everything, no frameworks, no dependencies, no build process
  • Live at a URL that anyone can use right now

What we learned

The hardest part of online impersonation is not detection, it's the workflow after. That gap is where harm happens, and it's where AI can help without replacing human judgment.

What's next for Takedown Copilot

Built With

Share this project:

Updates

Submission history