Elon Musk's X now trains Grok on your data by default – here's how to opt out

Generative AI tools have raised concerns over the way companies train them using everyone’s data, often without permission. Now you can add Elon Musk’s Grok AI to the list.

Earlier this week, Musk proudly announced that Grok had started supercluster training, added some technobabble about the effort, and then called it “the most powerful AI training cluster in the world.” Musk failed to reveal that Grok is also being trained on X users’ data, without their explicit permission.

Also: I tried X’s ‘anti-woke’ Grok AI chatbot. The results were the opposite of what I expected

The move came to light on X itself, posted by “EasyBakedOven,” or Kimmy Bestie, as spotted by Windows Central. In the post, EasyBakedOven said X had just activated a default setting for everyone that gives the platform the ability to train Grok on your data, but that this was never announced to users.

On its website, X describes the setting as such: “To continuously improve your experience, we may utilize your X posts as well as your user interactions, inputs, and results with Grok for training and fine-tuning purposes. This also means that your interactions, inputs, and results may also be shared with our service provider xAI for these purposes.”

You can’t disable the option from the mobile app, but you can from the website. To do so, open X in your browser. Select More at the bottom of the left pane and then click “Settings and privacy.” In the Settings section, click “Privacy and safety.” Scroll down to the Privacy and safety section and select Grok. Next to the setting “Allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning,” uncheck the checkbox.

If you’ve ever used Grok and want to erase all traces of your chats, click the link for “Delete conversation history” and then select Delete to confirm the action.

X isn’t alone in scooping up user information to train its AI models. Companies such as Meta and OpenAI also train their AI chatbots on publicly accessible user data. Why? The goal of generative AI is to sound and act like human beings. The more that AI models can learn from humans’ actual postings, conversations, and other interactions, the closer they come to achieving that goal.

The tricky part is when platforms like X enable this type of data collection by default without your knowledge or permission. The onus is then on you to learn which companies are collecting your data, how they’re using it, and how to prevent them from doing so.