Social networks have strengthened terms of use for scrapers and bots that crave websites to train AI models. A few days after X, owned by Elon Musk, updated its term and explicitly banned AI model training, today's decentralized social network Mastodon has updated its own rules banning all kinds of model training.
“We explicitly prohibit user data scraping for fraudulent purposes, such as archival models or large-scale language model (LLM) training. We want to make it clear that LLM for mastodon users' data is not permitted.”
New terms that will be applied to social networks starting July 1 include legal languages that prohibit the data extraction and development of automated systems.
“We use, launch, launch, develop and distribute automated systems that collect data mining or similar data mining or similar data mining or similar data mining or similar data mining or similar data mining or similar data mining or similar data mining or similar data mining or similar data mining to access instances, except in each case, including spiders, robots, cheat utilities, scrapers, offline readers, or data mining or similar data collection and extraction tools, for interaction with standard search engines or internet browsers, local caches, and the contents of the instances.
It is important to note that these terms only apply to Mastodon.social Server, which is just one instance of Fediverse, a distributed network. This means that the scraper can extract data from other servers and use it to train the AI model if the AI model does not explicitly ban it from service.
Other platforms, including Openai, Reddit and Browser Company, have added similar clauses to the rules to prevent other companies from training their models.
Apart from this change, Mastodon has implemented a new age limit of 16 for users. Social network age limits were 13 for users in the US, but the age limit has been changed worldwide.