X training Grok AI on user tweets without notification

X (formerly Twitter) has quietly enabled a setting that allows the platform to train its Grok AI on user tweets without notifying users.

The social network has opted all users in by default, using their posts, interactions, inputs, and results to refine Elon Musk's Grok AI model. This change means that users' data can be shared with the service provider xAI for training purposes.

To turn off this setting:

1. Open X on a web browser and log in.
2. Click on "More."
3. Choose "Settings and privacy."
4. Click on "Privacy and safety."
5. Select "Grok."
6. Uncheck the setting allowing Grok to use your content.

Currently, this setting can only be changed via the desktop version of X, though a mobile option is expected soon.

Users have expressed frustration over the lack of transparency. Recent updates to X's Help Center indicate the scope of data collection has expanded to include user posts explicitly. Initially, the policy mentioned using user interactions and inputs for training; now, it includes posts as well.

Purpose of Data Collection:

• Enhance Grok's language understanding.
• Improve response accuracy and relevance.
• Develop Grok's humor and wit.
• Ensure political neutrality and balanced answers.

Elon Musk recently highlighted that Grok is being trained on "the most powerful AI training cluster in the world" and is available to X premium subscribers.