2024-07-29 03:20:00
Several users of platform X, bought as Twitter by tycoon Elon Musk in 2022, noticed at the end of last week that the company changed without warning the data settings to allow itself to automatically train its new artificial intelligence (AI) model with all tweets sent to the network. Grok, as the intelligent machine that Musk promotes for X is called, would train this way with the posts of all its users unknowingly, unless they explicitly prevent that possibility in the…
Subscribe to continue reading
Read without limits
Several users of platform X, bought as Twitter by tycoon Elon Musk in 2022, noticed at the end of last week that the company changed without warning the data settings to allow itself to automatically train its new artificial intelligence (AI) model with all tweets sent to the network. Grok, as the intelligent machine that Musk promotes for X is called, would train this way with the posts of all its users unknowingly, unless they explicitly prevent that possibility in the platform settings.
This policy change, without explicitly notifying the tweeters, has raised eyebrows among privacy experts and online law specialists. Just two weeks ago, the European Commission opened proceedings against X for “deceiving users” with its use of the account verification system. This unexpected modification could trigger renewed scrutiny from regulators. A spokesperson for the Irish Data Protection Commission has already stated that they are “surprised” by this move and are awaiting a response from Musk’s team. Generally, EU data regulations do not allow the use of default-checked boxes or any other method of pre-set consent.
The way to deny X permission to collect data from each user’s posts is very simple, now that it has been revealed. It is enough to go to the Grok settings (at this link or from Settings > Privacy and security > Grok) and uncheck the option that says: “Allow your posts, as well as your interactions, entries, and results with Grok, to be used for training and improvement.” However, it can only be disabled in the web version, as the option is not available in the mobile app, which could certainly cause more problems for Musk’s legal teams.
In the additional information of that setting, X states that “to continuously improve your experience, we may use your posts on X, as well as your interactions, entries, and results as a user with Grok for training and improvement purposes.” It adds: “This also means that your interactions, entries, and results may be shared with our service provider xAI for those purposes.” In other words, they will share those databases with xAI, the company created by Musk to compete with OpenAI, the creator of ChatGPT.
Although it is not clear when X started harvesting user-generated content to train its AI model, at this time the platform is acting as if everyone had given default consent to improve the Grok tool. This AI model, Musk’s answer to the success of ChatGPT, was launched in November 2023 as a “rebellious” version that would indeed respond to “naughty” questions. Initially, the chatbot was not trained with X data, according to the company’s own specifications.
For the moment, only X’s paying subscribers have the ability to interact with Grok and conduct searches with its help. This means that currently the platform’s premium users are benefiting from the posts of all others. Paradoxically, when acquiring Twitter, Musk claimed that the “lords and serfs” system must end and tweeted: “Power to the people.”
Musk’s complaints about the elitist system of Twitter referred to the blue check verification, a mark that validated certain accounts alongside their name. It was the method the platform used to verify certain relevant or sensitive accounts that were at risk of being impersonated. In this way, all users knew that this person or organization was who they claimed to be. The tycoon decided to dismantle that system and convert it into a paid model, which opened the door to confusion: fake accounts appeared to be verified simply because they had spent a few euros in exchange for the blue check.
That is precisely what led the European Commission, on July 12, to publish a preliminary opinion warning X that it violates the Digital Services Act (DSA) “in areas related to dark patterns, advertising transparency, and access to data for researchers.” Brussels believes that X’s new account verification policy “does not correspond to industry practice and deceives users” and also considers that the social network does not meet the requirements for transparency and accountability concerning content moderation and advertising.
You can follow EL PAÍS Tecnología on Facebook and X or sign up here to receive our weekly newsletter.