French prosecutors suspect Musk encouraged deepfakes row to inflate X value

Elon Musk, Tesla CEO, attends the opening of the Tesla factory Berlin Brandenburg in Gruenheide, Germany, March 22, 2022. The intrigue surrounding Musk's Twitter investment took a new twist Tuesday, April 12, 2022, with the filing of a lawsuit alleging the colorful billionaire illegally delayed disclosing his big stake in the social media company so he could buy more shares at lower prices. (Patrick Pleul/Pool Photo via AP, File)

Warning: Graphic content.

PARIS - French prosecutors said Saturday they had alerted U.S. authorities to a suspicion that tech tycoon Elon Musk had encouraged the controversy over sexualised deepfakes on X to “artificially” increase the value of his company.

The social media network’s Grok AI chatbot stirred outrage earlier this year over it generating images of women and girls in minimal attire without their consent.

“The controversy sparked by sexually explicit deepfakes generated by Grok (X’s AI) may have been deliberately generated in order to artificially boost the value of companies X and X AI,” the Paris prosecutor’s office said, confirming a report in Le Monde newspaper on Friday.

This could have been done towards “the planned June 2026 stock market listing of the new entity created by the merger” between Space X and X AI, it added.

The prosecutor’s office said it had on Tuesday reached out to the U.S. Department of Justice, as well as French lawyers at the U.S. Securities and Exchange Commission (SEC), a financial market regulation body, to share its concerns.

X’s lawyer in France was not immediately available for comment.

French authorities are investigating X over allegations that its algorithm was used to interfere in French politics, as well as Grok’s dissemination of Holocaust denials and the sexualised deepfakes.

AI chat bot Grok has its own account on the X social network allowing users to interact with it.

For a period, users could tag the bot in posts to request image generation and editing, receiving the image in a reply from Grok. Many sent Grok photos of women or tagged the bot in replies to women’s photo posts, giving it prompts such as “put her in a bikini” or “remove her clothes”.

‘Incitements’

It generated an estimated three million sexualised images -- mostly of women, though also 23,000 that appeared to depict children -- in 11 days, the Center for Countering Digital Hate (CCDH), a nonprofit watchdog, said in late January.

Le Monde pointed to “several posts by Musk, published at the height of the controversy, which prosecutors interpret as incitements to generate non-consensual images”.

“The billionaire posted several messages in which he expressed delight, using numerous emojis, about his AI engine’s “undressing” capabilities, even sharing an image of himself in which his chatbot depicted him wearing a bikini,” Le Monde reported.

Daily average app downloads for Grok worldwide soared by 72 per cent from January 1 to January 19 compared to the same period in December, the Washington Post has cited market intelligence firm Sensor Tower as saying.

French authorities last month summoned Musk to a “voluntary interview” and searched the local offices of his social media network, in what Musk called a “political attack”.

Both Britain and the European Union have also opened investigations into the creation of the sexualised deepfakes.

By Clara Wright