IdeaWell

Project ideas from Hacker News discussions.

Show HN: Weedout – Safari extension that hides YouTube AI-labeled videos

Original Article

Hacker News Discussion

📝 Discussion Summary (Click to expand)

Theme 1 – AI labels are too vague or overbroad
- “Yep, it's super useless. The label is so general. Was AI used for one image in the 10 minute video? Is the voiceover AI? Is the entirety of it generated with AI? No way to know.” – f3408fh
- “Agreed. I think instead of 'AI or not', that something more nuanced is needed - perhaps a rating system inspired by the MPAA's (g, pg, pg-13 r, nc-17, x) that is used for movies.” – AmazingEveryDay

Theme 2 – Demand for better filtering/blocking tools (extensions, local models, cross‑platform)
- “It's free on GitHub for anyone to build, use or fork. I've built it for myself, but going through App Store submissions took me 2x the time, hence the price.” – masteranza
- “This is great, but why start with Safari and not a cross platform browser? Is there something native here that can't be done otherwise or is it just personal preference?” – lukeasch21
- “I’d love a small local model that detects AI voices and blocks the content.” – ungreased0675

Theme 3 – AI‑content filtering may exacerbate inequality (tech‑savvy can block, others cannot)
- “This is what they mean when they say AI amplifies inequality. If you're technically competent, you just write yourself an extension to block the slop spam, or at least pay to install one. So in the end it's rotting much more the brains of the least prepared.” – negura
- “I've built it for myself, but going through App Store submissions took me 2x the time, hence the price.” – masteranza (highlighting the cost/barrier for non‑technical users)

🚀 Project Ideas

AI‑Spectrum Browser Extension

Summary

  • A cross‑browser extension (Chrome, Firefox, Safari) that adds a nuanced AI‑involvement rating (0‑5) to videos, images, and audio on any site, combining platform labels, lightweight heuristics, and community feedback to overcome vague “AI” tags.
  • Core value: gives users granular control to filter, highlight, or trust AI‑generated media based on how much AI was actually used.

Details

Key Value
Target Audience Power users, researchers, and anyone frustrated by blunt AI labels on YouTube, TikTok, news sites, etc.
Core Feature Real‑time overlay showing a detailed AI‑usage score and breakdown (voice, visuals, music) with optional auto‑hide/filter thresholds.
Tech Stack WebExtension API, TypeScript, React + Redux for UI, TensorFlow.js for lightweight heuristic model, IndexedDB for user preferences.
Difficulty Medium
Monetization Revenue-ready: Subscription $2/mo for premium analytics and custom rule engine.

Notes

  • HN users asked for “this but for every website and media surface” (thatmf) and wanted “something more nuanced… a rating system inspired by the MPAA” (AmazingEveryDay).
  • Encourages discussion on what constitutes meaningful AI disclosure and can evolve into a community‑driven labeling standard.

VoiceGuard Local AI Voice Detector

Summary

  • A lightweight desktop/mobile app that runs an on‑device ML model to detect AI‑synthesized speech in audio streams and can mute, flag, or replace offending segments in real time.
  • Core value: privacy‑first solution for users who want to block AI voices without sending data to the cloud, addressing the pain point of annoying AI‑generated narration.

Details

Key Value
Target Audience Podcast listeners, video viewers, and accessibility‑focused users who dislike AI voices (e.g., karim79, ungreased0675).
Core Feature Real‑time AI voice detection using a small Whisper‑based classifier, with user‑configurable actions (mute, overlay warning, replace with silence).
Tech Stack Electron (desktop) / React Native (mobile), PyTorch Mobile model converted to ONNX, Rust audio processing core, optional Node.js backend for model updates.
Difficulty Medium-High
Monetization Revenue-ready: One‑time $5 for Pro model with higher accuracy and language packs; basic version free/Hobby.

Notes

  • Directly satisfies the request: “I’d love a small local model that detects AI voices and blocks the content.” (ungreased0675)
  • Sparks conversation about on‑device AI ethics, model size trade‑offs, and user empowerment over algorithmic content.

AI‑Metadata Hub for Creators

Summary

  • A service where creators can precisely tag which parts of their work used AI (voice, visuals, music, editing) via a simple UI or API; the hub emits a standardized JSON‑LD snippet that sites can embed, allowing browsers/extensions to show detailed, component‑level AI labels.
  • Core value: replaces vague “AI” tags with transparent, granular metadata, reducing false positives and enabling users to filter by specific AI components.

Details

Key Value
Target Audience Content creators (YouTubers, TikTokers, indie developers) and platforms seeking reliable AI disclosure.
Core Feature Creator dashboard to assign AI usage percentages per media component; API to generate and serve machine‑readable metadata tags.
Tech Stack Node.js/Express backend, PostgreSQL database, React frontend, JSON‑LD schema, optional IPFS for decentralized tag hosting.
Difficulty Medium
Monetization Revenue-ready: Tiered SaaS (free tier for limited tags, $10/mo for unlimited tags & API access).

Notes

  • Addresses the frustration that “the label is so general… Was AI used for one image in the 10‑minute video?” (f3408fh) and the desire for a “more nuanced” system (AmazingEveryDay).
  • Provides a concrete foundation for extensions like AI‑Spectrum to consume reliable data, fostering ecosystem‑wide better AI transparency.

Read Later