hschmale, 20 days ago

But it's a struggle to come up with a solution. Everything I can think of hurts accessibility of my content. Invisible characters could break screen readers. Dynamic #javascript or decrypt my content is bad, and isn't really that strong, so I'm at a loss.

ParadeGrotesque, 20 days ago

@hschmale

I have been asking myself the same question.

I think the answer is going to be:

• regularly updated 'robots.txt' file (found a couple that seem up to date).
• poison pill semantic nonsense using the META tag (so invisible to screen readers and most browsers).
• transient document space with more poison pill fake pages to lead the AI slurper to dead ends.
• some sort of copyright / copyleft notice squeezing very clearly "Not allowed to build AI corpus" or some such.

hschmale, 20 days ago

@ParadeGrotesque got any links for an up to date robots.txt?

Here's mine:
https://www.henryschmale.org/robots.txt

I'm thinking I might add a html comment to all my blog posts about that copyright/copy left.

But idk what to do beyond that.

And meta tags are still enough to do an awful lot and I need them to be good to make sharing easy.

ParadeGrotesque, 20 days ago

@hschmale

Here is a fairly up-to-date one: https://github.com/ai-robots-txt/ai.robots.txt