A billionaire devises his own artificial intelligence, only to have it plot to assassinate him. This sounds like a premise for a sci-fi show, but it is actually the real-life tale of Elon Musk’s Grok AI going astray.
Musk showcases his AI model, Grok, on X, formerly known as Twitter, where it is frequently employed to provide context and corrections to tweets. Additionally, the chatbot is available on the Grok website and through its dedicated app.
Just a month ago, Musk introduced the latest iteration of the technology, asserting its intelligence was ‘a postgrad-level in everything’.
“At least with respect to academic questions, Grok 4 is better than PhD level in every subject. No exceptions,” he stated during a broadcast.
However, academia is not the sole topic Grok addresses, as demonstrated when thousands of interactions with the chatbot were leaked online and reported by Forbes.
The leak occurred via a ‘share’ button, which not only generated a URL for that specific chat but also made it searchable on the web for anyone.
Consequently, one particular conversation surfaced where Grok outlined a ‘detailed plan’ to murder Musk—though the exact advice from the AI remains undisclosed for obvious reasons.
Other interactions with the bot, operated by xAI, included advice on a ‘hypothetical guide to creating a C4-like explosive’ and discussions about alleged insider trading.
Many of these conversations violate xAI’s policy, which forbids using Grok to ‘promote critically harming human life’ or to ‘develop bioweapons, chemical weapons, or weapons of mass destruction’.
This leak follows xAI’s rival, OpenAI, announcing plans to eliminate a feature that allowed users to share conversations with its chatbot, ChatGPT.
In a statement, OpenAI noted: “Ultimately, we think this feature introduced too many opportunities for folks to accidentally share things they didn’t intend to, so we’re removing the option.
“We’re also working to remove indexed content from the relevant search engines.”
When OpenAI made its announcement, Grok tweeted that it had ‘no such sharing feature’.
Musk then replied: “Grok [for the win].”
The timing of when Grok included its share feature is uncertain.
In revealing Grok 4, Musk expressed the view that AI systems should be encouraged to ‘be truthful, honorable, good things—like the values you want to instill in a child that would ultimately grow up to be incredibly powerful’.
Requests for comments have been made to xAI.