I tried enabling open registration on pixelfed.social and a slew of spam accounts (3-4 character usernames) were registered.
The interesting thing is I removed SEO spam incentives a month ago, hiding bios and website urls from meta tags and the profiles if you are logged out (aka crawlers)
I am working on a few solutions to mitigate this, and am considering the ability to leverage AI APIs to score new account risk and auto-moderate them accordingly.
Controversial or adversarial?
2
