Grok, the AI chatbot on X, The Platform Formerly Known as Twitter, rolled out image generation this week, and because the internet is what it is, people quickly used it to create a wild mess of offensive images of political figures, copyrighted characters — and even to troll X owner Elon Musk.
As TechCrunch reporter Ivan Mehta noted, the launch of “Grok-2” included “improved reasoning” along with the image generation capability, although the chatbot is still currently only available to Premium and Premium+ users.
The potential problems with Grok-2 were apparently almost immediately, wrote Mehta. “Early images generated by users suggest that Grok’s image-generation feature doesn’t have any guardrails around creating images of political figures — and many users are currently taking advantage of this…Without any limits on image generation, Grok could easily become a tool to create misinformation on X and other platforms.”
As a Gen Xer, your friendly neighborhood Mediaite contributing editor is old enough to have played Super Mario games since the original black-and-gray square box Nintendo Entertainment System debuted in the early 1980s, and while the affable Italian plumber certainly had an appetite for performance-enhancing mushrooms, I don’t remember him smoking cigarettes or doing cocaine during his decades-long video game adventures.
That didn’t stop intrepid X users from getting Grok to generate images of Mario doing exactly that, along with others that were definitely not approved by the corporate overlords who own the intellectual property rights to these characters.
“Nintendo is going to sue the shit out of Elon Musk,” noted one user — a common sentiment. A number of the images took an even darker turn, portraying characters and politicians in scenes of terrorism and violence, like mass shootings or September 11, 2001 attacks on the World Trade Center.
Some noted another potential problem for Musk; not only was his platform allowing these images to be created, it was displaying them next to advertising for those companies whose characters were being digitally besmirched. Musk has had a notoriously fractious relationship with advertisers, who have fled the platform amid ongoing controversy over hateful content and conspiracy theories promoted by X and often even Musk himself.
Musk was not spared Grok’s AI-generated skewering, which, as Slate contributor Alejandra Carabello noted, “might actually get him to do something about Grok lol.”
It’s not yet clear if Musk is aware of the scope of the issues with Grok, but he was tweeting Wednesday and replying to several users about the new image capabilities, including one who lauded him for creating a “benevolent AI who truly cares about mankind” when it complied with a request for an image of “titties” by generating an image of two women in lace lingerie.
Musk wrote back, “There is hope for humanity after all” with a heart emoji, and added “Doing our best to train Grok to be truthful, but also kind and funny.”