The TOS email says "We explicitly prohibit the scraping of user data for unauthorized purposes, e.g. archival or large language model (LLM) training. We want to make it clear that training LLMs on the data of Mastodon users on our instances, is not permitted."
Presumably this only applies to mastodon.social itself. I mean, if your post federates to Threads it's probably getting scraped for AI training over there.


