.jpg?w=1280&resize=1280,670&ssl=1 "Steve Bannon Is Out of Prison and Spreading Lies Online")
X collects user data for “Grok” AI training without consent
X (formerly Twitter) has introduced a feature that allows the company to use user data to train its AI assistant “Grok”.
The setting is enabled by default for all users and was implemented without prior notice, according to X users who found the setting. It allows X to use user interactions, inputs, and outputs with Grok for training and fine-tuning purposes. The collected data can also be shared with Elon Musk’s AI company xAI, which trains the models powering Grok.
Users can disable this setting on the website, but the option is hidden, and the mobile app doesn’t offer a way to opt out. On the web version, this link leads you to the “Grok Settings” option.
It might not bother Elon Musk, but this data collection likely conflicts with European data protection laws. Meta recently paused plans for its “Meta AI” assistant in Europe following objections from Irish data protection authorities.
Ad
Meta’s issue also involved using European user data for AI training, with automatic opt-in and a complicated opt-out process. Privacy advocacy group Noyb, which initiated the case against Meta, sees a clear violation of GDPR in using public user data for AI training without explicit consent, based on “legitimate interest”.
Meta expressed frustration with the decision, arguing that without European data, they could only offer an inferior product not optimized for local cultural nuances. The company also stated that Meta’s future multimodal AI models won’t be released in Europe because of “regulatory uncertainties.”
X’s actions will likely intensify the debate about using user data to train AI systems. Grok 2 is scheduled for release in August, with Grok 3 following in December. Musk claims Grok 3 will be the most powerful AI ever created.
