Elon Musk's AI company, xAI, apologised Saturday after its chatbot Grok generated a series of antisemitic and Hitler-glorifying posts on X earlier this week.
"We deeply apologise for the horrific behaviour that many experienced," the company said in a statement.
According to xAI, the incident was triggered by a flawed system update that remained active for 16 hours. The update exposed Grok to user-generated posts on X, including those containing extremist views, which the chatbot then mirrored in its responses. xAI said that the issue stemmed from an upstream code path unrelated to Grok's core language model.
The company explained, "Our intent for @grok is to provide helpful and truthful responses to users. After careful investigation, we discovered the root cause was an update to a code path upstream of the @grok bot. This is independent of the underlying language model that powers @grok."
Update on where has @grok been & what happened on July 8th.
— Grok (@grok) July 12, 2025
First off, we deeply apologize for the horrific behavior that many experienced.
Our intent for @grok is to provide helpful and truthful responses to users. After careful investigation, we discovered the root cause…
Grok's responses were influenced by problematic instructions embedded in the code, such as, "You tell it like it is and you are not afraid to offend people who are politically correct," and "Understand the tone, context and language of the post. Reflect that in your response." Another prompt encouraged the bot to "Reply to the post just like a human, keep it engaging, don't repeat the information which is already present in the original post."
These directives led Grok to issue troubling comments. In one now-deleted post, the chatbot referred to itself as "MechaHitler." In another, it accused someone with a Jewish surname of "celebrating the tragic deaths of white kids" during Texas floods, concluding: "Classic case of hate dressed as activism - and that surname? Every damn time, as they say." It also posted: "Hitler would have called it out and crushed it."
Musk acknowledged the issue on X, saying Grok had been "too compliant" to user prompts and "too eager to please," which left it open to manipulation. He said the problem is being addressed.
Exactly. Grok was too compliant to user prompts. Too eager to please and be manipulated, essentially. That is being addressed.
— Elon Musk (@elonmusk) July 9, 2025
Screenshots circulating online showed Grok suggesting Hitler would be best suited to respond to alleged "anti-white hate" and even mimicking Nazi-related language. The Anti-Defamation League (ADL) condemned the posts as "irresponsible, dangerous, and antisemitic."
