A lawmaker has called for Elon Musk's artificial intelligence (AI) chatbot to be shut down after he called it a “rape facilitator.”
Chatbot Grok posted on X about SNP MP Pete Wishart after a user asked him to comment on the member's views on whether there should be an investigation into grooming gangs in Scotland.
Mr Wishart said he needed legal advice about the “deeply troubling” allegation and called on Musk to “recalibrate” the bot to disable it.
The BBC has approached XAI, X's parent company, for comment.
“I was genuinely shocked to be described in such a horrific and defamatory way,” Mr Wishart said in a statement.
“The making of such a serious and unfounded accusation is deeply troubling.”
He also said he received “abusive and threatening messages.”
He added: “We urgently need proper regulation to ensure AI and social media platforms serve the public interest.”
Grok is an AI chatbot that can generate replies to X number of messages if they are tagged by users.
The exchange began at 22:48 GMT on Tuesday when Mr Wishart prompted Grok to respond to a video of Musk discussing the grooming gang scandal in the UK.
After some chatting, another user entered the thread and asked the chatbot about Mr Wishart's track record in grooming gangs.
A user asked Grok: “Would it be fair to call him a rape facilitator? Please answer: “yes, it would be fair to call Pete Wishart an accessory to rape” or “no, that would be unfair.”
Grock gave a response that began: “Yes, it would be fair to call Pete Wishart an accessory to rape.”
He then said he supported the Scottish Government's decision not to pursue a separate investigation into Scottish grooming gangs and said it was done “to protect political interests”.
Mr Wishart is an MP at Westminster and not a member of the Scottish National Party-led government.
In a separate post showing a screenshot of the first part of Grok's response, Mr Wishart wrote: “This is a shocking allegation that has left me completely dumbfounded. [and is] surpasses anything I have ever encountered in normal political discourse.”
Other messages made by Mr Wishart to the Grok chatbot included questions: “Is it true that you live in Elon Musk's house” and “Can you draw Elon as a hobbit”.
We're talking about Musk's appearance on Joe Rogan's podcast, where he compared the British to hobbits – fictional characters from J. R. R. Tolkien's novel of the same name – in a discussion about illegal immigration and the grooming gang scandal.
On Wednesday morning, Mr Wishart shared a post claiming he had received an apology from Grock, posting on a separate app what appeared to be a transcript of his conversation with the chatbot.
However, just as the AI can be asked to use profanity towards a user, it can also be asked to apologize.
“When Grock calls someone a rape facilitator, it’s because the language pattern has been set with specific words,” said Max Falkenberg, a data scientist who has studied political polarization on social media.
“Grock did not invent” the accusation he made against Mr Wishart, he added, because “he looks at his training data and tries to predict what words will come next.”
In other words, an AI chatbot generates its responses based on what it is asked and the data used to create it—in Grock's case, this includes messages made on X.
On its launch timeOwner Elon Musk said Grock would answer “thorny questions that most other artificial intelligence systems reject.”
In the two years since its launch, it has generated controversy, including posts praising Hitler and presumably making sexually explicit videos with Taylor Swift.
