Several tools aim to protect intellectual property, keep AI from learning from your content, and confuse tracking by making collected data hard to interpret or misleading.
1. Protect your writing: ShieldFont
- ShieldFont is an open-source web font. It swaps words in a page’s source with other real words, so readers still see your original text while scrapers collect the substituted version.
- Shielded passages are hidden from screen readers by default, with a control that reveals the real text.
- The aim is to keep AI models from scraping and learning from your written ideas.
2. Protect your images: Nightshade and PhotoGuard
- Nightshade, from the University of Chicago, subtly alters images so that models trained on them without permission learn the wrong things, while the images look the same to people. The goal is to make unlicensed training costly.
- PhotoGuard, a research prototype from MIT, adds invisible changes that make AI-powered editing of a photo much harder.
3. Protect your browsing privacy: TrackMeNot and AdNauseam
- TrackMeNot sends randomized search queries in the background to hide your real searches.
- AdNauseam blocks ads and quietly clicks them, so ad networks get a click trail that says nothing about you.
- Both make it harder for companies to infer your interests, spending habits and psychological traits.
Key takeaway
Instead of only hiding data from AI, these tools make collected data less useful or less trustworthy. Widespread use could undermine AI training, surveillance and profiling.
Caveat: these tools work in different ways and have limits. None of them guarantees protection against scraping, training, manipulation or tracking.
Related: AI and decisions · Decide what AI must never do