X blocks companies from training AI models on its posts
X just updated its developer agreement to bar third parties from using platform posts or its API to train or fine-tune foundation models. The restriction mirrors Reddit’s playbook and could pave the way for lucrative data licensing deals. Meanwhile, X is still freely feeding user data into its own Grok AI.
The era of casually scraping X to train frontier models is officially over.
According to an update spotted by TechCrunch, X has revised its developer agreement to explicitly ban anyone from using platform content or its API to fine-tune or train foundation or frontier models.
Closing the gates to open a tollbooth
Don't mistake this for a total data lockdown. While everyday developers are locked out, X's privacy policy still lets authorized third-party collaborators train on user data unless users manually opt out.
And then there is family privilege. Elon Musk’s AI company, xAI, which acquired X on paper for $33 billion in March, continues to feed platform data directly into its own Grok model.
Why it matters
This looks like a classic playbook move to turn free social chatter into high-margin licensing revenue. Reddit did the exact same thing before landing a lucrative data deal with Google, and Reddit just sued Anthropic on Wednesday for allegedly crawling its site over 100,000 times since July 2024.
For AI startups and researchers, one of the web's biggest real-time dialogue archives is now behind an enterprise wall. We think frontier labs will have no choice but to cut massive checks for official data access, cementing a pay-to-play training landscape where independent builders are simply priced out.
Sources (1)
- theverge.com(opens in a new tab)Originaltheverge.com