Project ideas from Hacker News discussions.

Google Chrome silently installs a 4 GB AI model on your device without consent

📝 Discussion Summary (Click to expand)

1. Massive storagebloat from Chrome’s AI add‑on > “4GB of storage is not a ‘mere’ thing, to the contrary.” – paganel

2. Users feel they are not asked for consent / forced into opt‑out

Maybe consent is not an appropriate term. Perhaps an acknowledgement and a way to say “I don’t want this” would be a more suitable approach.” – scorpioxy

3. Distrust of Google’s “AI‑everywhere” strategy

I think they do. They just don’t care.” – delecti

4. People are migrating to alternative browsers / forks

I use Firefox.” – chinathrow

5. Memory / performance concerns differentiate Chrome from rivals

Chrome also came in at slightly lower memory consumption … 4.67 GB vs 4.83 GB for Firefox.” – theandrewbailey

6. Skepticism about the purpose of the locally‑installed model

The model doesn’t do anything, it’s just a sandbox.” – jauntywundrkind

7. Privacy worries – fear of data being shipped back to Google

I certainly would not trust Google or Chrome. They have a history of class‑action lawsuits for doing shady things to users.” – 2ndorderthought

8. Real‑world impact on metered or low‑bandwidth users

Places like public school districts have hundreds of computers and this will be somewhat significant for them.” – dd_taylor


🚀 Project Ideas

Opt-OutGuardian for Browser AI Downloads

Summary

  • Detects and flags unexpected large file downloads (e.g., AI model weights) by browsers without explicit user consent.
  • Provides a one‑click toggle to block or allow such downloads, respecting OS metered‑network settings.

Details

Key Value
Target Audience Privacy‑conscious power users, developers, and enterprise IT teams
Core Feature Real‑time monitoring of Chrome/Firefox/Edge download activity, integration with OS network‑metered APIs, user‑controlled allowlist
Tech Stack Electron front‑end, Node.js background service, WebExtension APIs, native OS hooks (Windows Metered, macOS NetworkQuality), SQLite storage
Difficulty Medium
Monetization Revenue-ready: $5/month

Notes- Directly solves the consent‑opting problem repeatedly voiced on HN.

  • Can be packaged as a browser extension or standalone utility, attracting both individual and corporate adopters.

Transparent Local LLM Manager

Summary

  • Gives users a clear dashboard to view, download, and manage locally cached AI models, showing size, source, and storage impact.
  • Allows selective enable/disable of model components on a per‑application basis.

Details

Key Value
Target Audience Developers, privacy‑focused users, and researchers experimenting with on‑device LLMs
Core Feature Integrated UI showing model provenance, download progress, disk usage heatmaps, and per‑model opt‑out toggles
Tech Stack React web UI, Electron wrapper, WebAssembly for model verification, backend in Rust, uses OS file‑system APIs
Monetization Hobby

Read Later