French authorities are investigating Elon Musk’s AI chatbot Grok after it generated posts denying the Holocaust at Auschwitz.
The Paris prosecutor’s office confirmed Friday that the remarks have been added to an existing cybercrime probe into X. The investigation highlights growing European concerns about AI-generated misinformation on large social platforms. France has some of Europe’s toughest Holocaust denial laws.
Grok, integrated into Musk’s platform X, wrote in French that the Auschwitz gas chambers were designed for “disinfection with Zyklon B against typhus” rather than mass murder — language long associated with Holocaust denial.
The Auschwitz Memorial flagged the exchange, saying it distorted historical facts and violated platform rules. Grok later acknowledged the error and deleted the post.
In follow-up messages, the chatbot pointed to historical evidence showing that Zyklon B gas chambers murdered more than 1 million people at Auschwitz. X has not issued any public clarification.
This is not Grok’s first antisemitic controversy. Earlier this year, Musk’s company removed posts from the chatbot that appeared to praise Adolf Hitler after complaints.
French Industry Minister Roland Lescure reported Grok’s posts to prosecutors under a legal requirement that public officials flag potential crimes. Government officials called the AI-generated content “manifestly illicit,” saying it could constitute racially motivated defamation and denial of crimes against humanity, according to the Associated Press.
Prosecutors said “the functioning of the AI will be examined” as part of the expanded investigation, which originally focused on whether X’s algorithm could be used for foreign interference.
French authorities have also alerted the country’s digital regulator over suspected violations of the EU’s Digital Services Act and referred the posts to a national police platform for illegal online content.
The European Commission said this week it is in contact with X about Grok, calling some of the chatbot’s output “appalling” and contrary to Europe’s fundamental rights and values.
Two French rights organizations — Ligue des droits de l’Homme and SOS Racisme — filed criminal complaints accusing Grok and X of contesting crimes against humanity.
Tests by The Dallas Express on Friday showed Grok providing historically accurate information about Auschwitz. When asked whether it had previously issued inaccurate content on the subject, Grok acknowledged the mistake, calling it a “classic and dangerous trope of Holocaust denial” and saying it “understandably sparked outrage.”
“To be clear, this output was a failure in my safeguards and training data handling — it doesn’t reflect truth or my core design to seek it,” Grok wrote in response.