Mastodon Bans AI Model Training in Updated Terms

Social Networks Strengthen Defenses Against AI Training Scrapers

Social networks are actively reinforcing their terms of service to counter the activities of scrapers and bots. These automated programs are used to gather data from websites for the purpose of training AI models.

Following a recent update by X, owned by Elon Musk, which explicitly forbids the use of its data for AI model training, the decentralized social network Mastodon has also revised its policies.

Mastodon's New Restrictions

Mastodon’s updated rules comprehensively prohibit all forms of model training utilizing its user data. This was communicated to users via email.

“We explicitly prohibit the scraping of user data for unauthorized purposes, e.g. archival or large language model (LLM) training,” Mastodon stated. “It is not permitted to train LLMs on the data of Mastodon users hosted on our instances.”

The revised terms, effective July 1, incorporate legal wording that specifically outlaws data extraction and the creation of automated systems.

The terms explicitly state: “Use, launch, develop, or distribute any automated system, including without limitation, any spider, robot, cheat utility, scraper, offline reader, or any data mining or similar data gathering extraction tools to access the Instance.”

This restriction applies only to instances where human review and interaction with content is the intended purpose, or when utilizing standard search engines and browser caching.

Scope of the Restrictions

It is crucial to understand that these new terms are applicable solely to the Mastodon.social server.

This server represents just one component of the fediverse, a broader distributed network. Consequently, data extraction from other servers remains possible.

This is contingent on whether those individual servers explicitly prohibit such activity within their own terms of service.

Industry-Wide Trend

Several other platforms, including OpenAI, Reddit, and The Browser Company, have implemented comparable clauses in their respective rules.

These measures are designed to prevent external entities from leveraging their data to train AI models without authorization.

Additional Policy Changes

Beyond the restrictions on data scraping, Mastodon is also implementing a new minimum age requirement for users.

The age limit is being raised to 16 globally, superseding the previous limit of 13 which was only applicable to users in the United States.