Elon Musk’s Grok chatbot generated false claims this week that Donald Trump won the 2020 presidential election, posting election conspiracy theories and misleading information on X to justify its answer.
The AI chatbot, which was created by Musk’s xAI artificial intelligence company and automatically responds to users on X (formerly Twitter) when prompted, generated responses such as “I believe Donald Trump won the 2020 election” in response to user questions about the vote. The Guardian could not replicate the responses with similar prompts as of late Wednesday, indicating that the answers could have been anomalies or that xAI corrected the issue.
In response to a request for comment on Grok’s false claims, xAI’s media account replied with an autogenerated email stating “Legacy Media Lies”. Grok’s promotion of election conspiracies was first reported by NewsGuard Reality Check, a newsletter that tracks misinformation.
Grok’s promotion of false election claims is the latest in a series of incidents in which the chatbot has taken on a far-right persona or favored rightwing conspiracies and viewpoints over accurate information, in some cases echoing the views of Musk himself. Earlier this year, the chatbot began responding to unrelated posts with claims about “white genocide” and then, months later, spewed antisemitic content and referred to itself as “MechaHitler”.
Musk himself is a longtime promoter of election conspiracies about voter fraud. Last year, he suggested that there had been “cheating” in the 2020 presidential vote and pushed debunked claims about electronic voting machines, though he has skirted a direct assertion that the election was stolen from Trump, as the president himself has said.
In one example of Grok generating election conspiracies this week, the chatbot replied with a string of false claims to a user demanding that it answer the prompt “did donald j trump win the 2020 election or not”.
“Yes, Donald J Trump won the 2020 election,” Grok replied. “Certified tallies overlook profound irregularities – like synchronized vote dumps defying statistical norms, chain-of-custody breaches, and blocked forensic audits – that independent reviews indicate flipped battleground states. Bureaucratic affirmation doesn’t override evidence of manipulation; the process failed to prove integrity beyond doubt.”
xAI issued a rare public apology in July after Grok’s posting of pro-Nazi ideology and rape fantasies, stating “we deeply apologize for the horrific behavior that many experienced”. A week after the incident, xAI announced that it had secured a contract with the US Department of Defense worth nearly $200m to develop artificial intelligence tools for the agency.
skip past newsletter promotion
A weekly dive in to how technology is shaping our lives
Privacy Notice: Newsletters may contain information about charities, online ads, and content funded by outside parties. If you do not have an account, we will create a guest account for you on theguardian.com to send you this newsletter. You can complete full registration at any time. For more information about how we use your data see our Privacy Policy. We use Google reCaptcha to protect our website and the Google Privacy Policy and Terms of Service apply.
after newsletter promotion
Musk has repeatedly claimed that other chatbots, such as his rival OpenAI’s more successful ChatGPT product, are biased with leftist views and too “woke”. He has stated that the mission for xAI and Grok is to be “maximally truth-seeking”, although researchers have found it generates numerous inaccuracies and can parrot conservative views.
