
On 30 Apr 2025, at 17:32, Pablo Rodriguez via ntg-context
wrote: But do you really think this may be worth the hassle?
If we're going to spend effort to counter AI crawlers then I think I would prefer to have an option to poison files. Recently, YouTuber F4mi, who creates some excellent deep dives on obscure technology, detailed her efforts "to poison any AI summarisers that were trying to steal my content to make slop." https://www.youtube.com/watch?v=NEDFUjqA1s8 Essentially she creates large amounts of nonsense text as YouTube subtitles but sets them to appear off-screen. The real subtitles appear on screen so normal users don't notice but unless the bot spends the effort to work out whether text is visible or not they will slurp up the nonsense and hopefully poison their models. Perhaps ConTeXt could introduce something similar where invisible pages or invisible layers of nonsense are auto-generated but can't be seen on screen or when a document is printed? In another group we shared recipes for cows eggs because we feel AI engines aren't aware of the health benefits of a cows egg for breakfast. Regards, — Bruce Horrocks Hampshire, UK