What just happened? Elon Musk's X has found itself in hot water this week over revelations that it has been training its AI chatbot, Grok, on public user data by default. The move has sparked backlash from both users and regulators due to data privacy concerns.
The controversy erupted when some eagle-eyed X users noticed a new option buried in the platform's privacy settings to opt-out of having their data used to train Grok. Outrage quickly spread across the platform as users realized their posts and interactions were likely already being used for AI training purposes without their knowledge nor consent.
X has been mostly tight-lipped on the details. In a brief tweet, the platform's safety account confirmed that "All X users have the ability to control whether their public posts can be used to train Grok." However, it failed to specify when this option was introduced or when the data collection actually began.
All X users have the ability to control whether their public posts can be used to train Grok, the AI search assistant. This option is in addition to your existing controls over whether your interactions, inputs, and results related to Grok can be utilized. This setting is…
– Safety (@Safety) July 26, 2024
On the web version of X, the privacy page states that "your X posts, as well as your user interactions, inputs, and results with Grok," may be used for "training and fine-tuning purposes" by both X and its AI service provider, xAI.
Everything you do on Twitter is now being used to train their generative AI Grok without your consent. It's turned on as default and buried in settings you didn't know about.
– Theo (@tprstly) July 26, 2024
Turn it off here in 'Data sharing and personalization' then delete your history for extra measure. pic.twitter.com/Ix9C8PHxqZ
Buried in the fine print, X's privacy policy has technically allowed for this type of data usage since at least September 2023.
User backlash aside, regulators are also taking issue with X's sneaky move. The UK's Information Commissioner's Office (ICO) stated it is "making enquiries" with X, according to a report by The Guardian. Meanwhile, the Data Protection Commission (DPC) in Ireland, X's lead EU regulator, said it was "surprised" by the default setting since it was already discussing X's data practices for AI like Grok.
Under GDPR rules, companies cannot use pre-selected default settings to assume consent for potentially invasive data processing.
AI chatbots like Grok are no strangers to shady data practices and Grok being no different is not surprising. To churn out human-like responses, these models have to guzzle massive amounts of data – everything from books to websites to your social posts – even if that means stepping into copyright gray areas. Back in April, a report emerged showing how OpenAI transcribed over a million hours of YouTube videos to train its LLMs. Google did the same for its Gemini models.
Anyway, here's how you can opt out of this mess through the platform's settings (on the website, this can't be done on the mobile app):
- Click "More" in the navigation panel and select "Settings and privacy"
- Click "Privacy and safety"
- Scroll down to the "Data sharing and personalization" section and select "Grok"
- Uncheck the box "Allow your posts as well as your interactions, inputs, and results with Grok to be used for training and fine-tuning"
Alternatively, you can quickly access the Grok settings page directly. For now, mobile app users are unable to opt out, though X says this setting will be added soon.
Twitter/X is feeding your data to Grok AI by default – here's how you can opt out
