Mastodon updates its terms to prohibit AI model training

In a bid to strengthen its defenses against unauthorized data use, Mastodon has revised its terms of service to explicitly prohibit the training of AI models using user data. This update comes closely on the heels of a similar move by X, the social platform owned by Elon Musk, which recently amended its own policies to ban AI model training. In an email to users, Mastodon clarified, "We explicitly prohibit the scraping of user data for unauthorized purposes, such as archival or large language model (LLM) training." The newly enforced rules, effective from July 1, include legal provisions that strictly forbid any extraction of data and the development of automated systems designed to access their platform. The updated terms detail that users are not allowed to utilize or create any automated tools, including spiders, robots, or scrapers, to collect data from the Mastodon instances. The only exceptions to this rule are standard search engine functions or local caching for human interaction with the content. It’s crucial to recognize that these new regulations specifically apply to the Mastodon.social server, which is merely one instance within the broader fediverse network. Consequently, scrapers may still be able to scrape data from other servers that do not have similar prohibitions in place. In addition to these changes, Mastodon is also implementing a new global age restriction, raising the minimum user age from 13 to 16. This policy shift aligns with the platform's commitment to creating a safer environment for its community.