A recent shake-up at Elon Musk’s X is now set in stone: an update to the platform’s terms of service allows it to gather user data to train its AI models.
If you’re still on the social media platform from November 15, X can feed your tweets, photos, and videos to its Grok AI chatbot.
That’s the digital helper that you probably don’t even use because it’s locked behind a monthly subscription. You know, the one X tried to get people to sign up to by removing blue ticks.
“By submitting, posting or displaying Content on or through [X]” the terms of service state, you’re granting it the right to “analyze text and other information you provide…for use with and training of our machine learning and artificial intelligence models, whether generative or another type”.
The change has repeatedly been flagged in the past year. Musk previously confirmed that his xAI firm, a budding rival to OpenAI valued at an eye-watering $24bn, would use public tweets to train Grok.
It was also listed on the “humourous” chatbot’s help page, which states: “To continuously improve your experience, we may utilize your X posts…for training and fine-tuning purposes.”
That might explain why Grok’s responses are so unapologetically blunt. After all, Twitter thrives on wisecracks and snappy one-liners designed to go viral.
In a recent example, the bot responded to a user’s prompt by branding Musk “one of the most significant spreaders of misinformation on X”.
X/Twitter users’ reactions to the AI policy
But, with the divisive policy now taking effect, users are threatening to leave X and encouraging others to follow suit. Creatives are particularly concerned about the contentious update as they use X to share their work.
“Because of the upcoming AI training effective tomorrow here, I will (sadly) be deleting all my posts here and will no longer be active,” posted comic book artist Rian Gonzales. “I'm sad because I've met so many of you incredible people here.”
The illustrator’s profile name announced her move to Bluesky, the X rival started by Twitter co-creator Jack Dorsey that has gained traction in recent weeks.
The latest TOS agreement says X has the legal right to use your info to train it’s AI even if you aren’t a paid subscriber. I'm downloading my archive and bailing out of here this week. I'd suggest you do the same. pic.twitter.com/gzFSFOU7sY
— Drew "bluesky" Schnoebelen (@Dschnoeb) November 12, 2024
Bluesky recently revealed that it had added more than one million new users since the US presidential election, benefiting from an exodus of those unhappy with X.
“The latest TOS agreement says X has the legal right to use your info to train its AI even if you aren’t a paid subscriber,” wrote Drew Schnoebelen, a Penguin Random House employee, whose profile also mentions Bluesky. “I'm downloading my archive and bailing out of here this week. I'd suggest you do the same.”
X isn’t the only platform fuelling its AI on user data. Meta recently implemented a similar policy for its social networks, including Facebook and Instagram. In a controversial move, the tech giant announced that users would need to submit a form to request removal, rather than being able to opt out by default.
Meanwhile, LinkedIn was forced to walk back its own AI data-sharing policy after it sparked a user uproar.
In addition, Reddit (another site where users are complaining about X’s revised rules) is allowing Google to access user content for AI training purposes in exchange for $60million in licensing fees.
Can you opt out?
At the time of writing, users peeved by the change can still opt out of having their data used for AI training - but it’s unclear how long this option will work.
Legally speaking, though, it may not matter. The terms of service document don’t explicitly mention an opt-out option, and it seems irrelevant whether your account is locked — based on these terms, Grok can still use your data.
Still, if you want to give it a go for peace of mind, here’s how to do it:
On X, open “settings and privacy” > then tap “privacy and safety” > now scroll down and tap “Grok” > and untick the box that states “allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning”.
What else is changing on X?
The AI stuff isn’t the only tweak keeping X users awake at night. As part of the new terms of service, X also states that disputes raised by users will be governed by “the laws of the State of Texas”.
All disputes, the terms state, “shall be brought exclusively in the US District Court for the Northern District of Texas, or state courts located in Tarrant County, Texas”.
The Washington Post recently reported that X's decision to modify its terms of service aims to steer lawsuits to courthouses that are considered a “hub for conservatives”. The move is believed to have been made by Musk to better shield X from potential litigation.
Although not mentioned in the new rules, X also recently altered its block feature so that blocked users can now view the public posts and profile page of the account that has blocked them.
Many users and experts have raised concerns about the update’s impact on online safety, warning that it may allow unwanted followers to continue monitoring users who have blocked them.
This change could especially affect high-profile individuals, particularly women, they noted, who are more likely to face online harassment.
