AI Honeypot Builder

AI bots that ignore instructions telling them not to scrape content piss me off. The goal of this project is to build a tool that builds sites to poison the bad actors. The idea being to deploy them on all those unused domains you have laying around or subdomains on existing sites.

Check the repo for updates: ai-honepot.alanwsmith.com repo

You can check out an example of the output here

Some Basic Ideas

      - Randomize number of pages, paragraphs, and sentences
      - Add different output templates
      - Add links to other pages on the home page 
      - Add cross links between pages 
      - Add analytics
      - Add some third party scripts
      - Add youtube embeds base of search strings that match the content
      - Add the occasional spelling error
      - Create links with actual titles to other sites
      - Make data tables with stats from sports
      - Use open source images with different labels
      - Manipulate images (see nightshade)
      - Pull in lists of celebrity names
      - Pull in song, move, show titles
      - Make them in different languages
      - Pull wikipedia pages and reassemble them in random ways
      - Use varying amounts and types of metadata
      - Create sites with different frameworks
      - Change directory structures randomly
      - Include CSS
      - Include code samples
      - cross linking between multiple sites
      - site maps
      - rss feeds
    

Notes

• These sites will only impact scrappers that ignore the directives in robots.txt. Legit scrappers and bots will safely ignore the contents by following the parameters of the site.
• The goal is not to stop AI scrappers. It's just to poison bad actors. If AI bots followed the rules this wouldn't be necessary. It would also mean the decisions of folks who don't want their stuff ingested by AI bots would be honored.
• The odds of enough folks doing this to make real impact are probably low. But, if none of us try, we'll never know
• I'd be shocked to learn there aren't other tools out there that do this. My thinking is, the more the better since they'll all take different approaches
• It probably wouldn't take much to counteract these sites. They'd have to be alerted to them first which could happen instantly or never. Either way, they'd have to expend resources to mitigate the impact