Mastodon Bans AI Training on User Data, Highlighting Challenges in Decentralized Platforms

Training AI on Mastodon posts? That idea's extinct

Such rules could be tricky to enforce in the Fediverse, though Mastodon is the latest platform to push back against AI training, updating its terms and conditions to ban the use of user content for large language models (LLMs). "We want to make it clear," the federated platform stated in an email to users, "that training LLMs on the data of Mastodon users on our instances is not permitted." The announcement may feel like shutting the stable door after the horse has bolted, but it's still reassuring to know that users' rants on the platform, in theory, won't feed into the LLMs behind generative AI services. To be fair, enforcing such restrictions on a platform that prides itself on decentralization and openness could prove difficult. The terms apply only to Mastodon's own instances, not the wider Fediverse. It's possible to deploy a file to block AI crawlers, but that relies on those behind the bots respecting it rather than invoking fair use. Mastodon is not the only platform worried about its content being used for AI training. Another social media platform, Bluesky, recently said: "We do not use any of your content to train generative AI, and have no intention of doing so," but, as the service acknowledged, enforcement of such a rule outside its systems is challenging. As 2024 drew to a close, a million public posts from Bluesky's firehose API turned up in a training set. Earlier in June, discussion forum Reddit sued Anthropic, an AI business, over allegations [complaint is here - PDF] that content generated by its users was scraped in violation of contractual terms and technical barriers. The suit did not cite examples of any alleged violations by Anthropic after July 2024. In 2024, Reddit signed a data-sharing deal with OpenAI. Earlier that year, it signed an AI training deal with Google, having begun charging companies to use its data-downloading API in 2023. Mastodon's change highlights the concerns of users over how their data might be used, particularly on platforms that are, by their nature, as free and open as possible. The updates, including an increase in minimum age from 13 to 16, take effect from July 1. ®

Mastodon's New Terms Block AI Scraping, But Gaps Still Remain

Open-source social media platform Mastodon has changed its rules to block companies from using user posts to train AI systems. The new policy, which takes effect from July 1, specifically bans scraping data from its main server, namely 'Mastodon.social'. The updated terms of service now clearly say that automated tools like scrapers or data miners cannot collect information from the platform. This includes collecting data for purposes like archiving or training large language models (LLMs), which power AI tools such as chatbots. The terms state to users that, "You are prohibited from using the Instance (the main server) for the commission of harmful or illegal activities." Accordingly, Mastodon instructs that no one should attempt, or assist any other person who wants to attempt, to access users' data from its main server. This essentially means that one cannot collect data from the 'Mastodon.social' server using bots or other automated tools, unless it is for normal internet browsing or human review, etc. Pushback Against AI Overreach The change comes amid growing anger among users of various platforms about their public posts being used without permission to train AI systems. Similar concerns were recently raised on other networks like Bluesky, where a dataset of user posts was collected and uploaded for research purposes. AI data scraping has become a wider issue in the tech world. Major platforms such as Reddit have also taken action. In one high-profile case, Reddit filed a lawsuit against AI company Anthropic, accusing it of using Reddit content without permission to train its Claude chatbot. Mastodon's updated policy may come as a relief to those who worry about how AI is vacuuming up the internet's content. However this change only applies to 'Mastodon.social' -- the main server run by Mastodon gGmbH, the non-profit entity behind the platform. Mastodon operates as a decentralised network, with an array of various independently managed servers. Therefore, if other server administrators don't adopt similar rules, users across the broader "fediverse" could still face exposure. Mastodon Raises Age Limit For Users In addition to the scraping ban, Mastodon has also raised the minimum age -- for users around the world -- from 13 to 16 years. The updated policy bars anyone under 16 from accessing the server and requires users who cannot legally accept the terms to have a parent or guardian accept them instead. More Legal Protections, But With Limits The social media platform also updated its terms to include stronger rules against hacking attempts, and automated tools that can disrupt the platform. If a user flouts these terms, Mastodon reserves the right to deny them access, and even report them to law enforcement authorities. The updated terms also include an arbitration clause, meaning users who want to sue must go through a legal process in Germany. This could make it harder for users in other countries to challenge Mastodon in court. And while the platform gives users ownership of their own posts, it grants itself broad rights to use and share that content. Not a Full Win for Users While Mastodon's efforts to push back against AI scraping are a welcome move for privacy-conscious users, the policy's limited scope -- applying to only one server -- leaves much of the "fediverse" unprotected. It also highlights how difficult it is for decentralised platforms to act as one unit against AI firms harvesting online data. Unless more servers adopt the same rules as 'Mastodon.social', AI companies may still gorge on user content to train datasets, with or without users' consent. Also Read: