The scandal reverberated far beyond Europe’s borders. Grok, the artificial intelligence tool of social network X, has been under fire from critics around the world for over a month. At issue is its built-in image generation function, which has enabled several thousand users to create sexually explicit deepfakes of women, sometimes even minors, without their consent (see EUROPE 13779/14).
Calls for the Commission to take action under the Digital Services Act (DSA) to which X is subject have multiplied since the beginning of January (see EUROPE 13782/7). On Monday 26th, the Commission finally announced an official investigation into the social network in order to “determine whether the company has properly assessed and mitigated the risks associated with deploying Grok’s features on X within the EU”.
“This includes the risks associated with disseminating illegal content such as manipulated sexually explicit images, including content that may constitute child sexual abuse material”, explains the institution.
The image generation function was initially restricted to paying subscribers only (see EUROPE 13783/11), before being removed for all users a few days later (see EUROPE 13787/7).
Under the DSA, platforms are required to carry out studies into the risks that some of the features they offer their users may pose in terms of gender-based violence, mental health or the dissemination of illegal content.
However, the integration of Grok into X, and therefore its open access to any user, aren’t mentioned in the platform’s recent reports, according to a senior European official. This absence leads the Commission to suspect that X has not taken the necessary steps to properly anticipate the risks for the general public that might be associated with this access, and has therefore failed to meet its DSA obligations.
The announcement of the investigation was widely welcomed by a whole section of Europe’s political class and civil society organisations. However, some continue to regret the late decision, which was too lukewarm given the controversy.
To counter these criticisms, the Commission insists that Grok has been under surveillance for some time (see EUROPE 13678/18). While the measures implemented by X have not ruled out the need for an investigation, they do avoid the need for interim measures, such as Article 70, which would have effectively suspended Grok in the EU and which had been called for by several MEPs (see EUROPE 13789/10).
In parallel with this new investigation, the Commission is expanding another investigation into X, opened at the end of 2023 (see EUROPE 13316/10). The EU suspects X’s algorithms of amplifying the dissemination of illegal content and spreading false information.
In the coming months, Grok should be used to read all the posts and watch all the videos posted on X “in order to offer users the content that is likely to be of most interest to them”, according to Elon Musk. Given this change and Grok’s previous issues, the Commission wants to know whether X has properly assessed and mitigated the potential risks. (Original version in French by Isalia Stieffatre)