Recently, we got the news that Grok-2, Elon Musk’s AI model gained the ability to generate images. This brings it up to speed with other AI models, and the images it produces are pretty realistic. However, there don’t appear to be any filters on it. Apparently, Grok is generating some reckless images.
The Grok platform is available to people subscribed to the X Premium+ plan on X.com. This is the $16/month ($22/month on mobile) payment tier for the social media site. It offers searching capabilities and other features. Along with that, you’ll all get other features. You can check out Everything You Need To Know About X Premium for more information.
Grok generates reckless images
Were we really surprised? Just about everything involving Elon Musk this year has been drenched in controversy. So, his Grok-2 model can generate images; what could go wrong?
According to several reports, this platform can make some decent-looking images. So, if you want to make a photo of a purple cat patting Gordon Ramsay on the head, go ahead! Obviously, if you want to generate something potentially brazen like former U.S. president Barack Obama stabbing Joe Biden, it would be blocked, right? Right!? Remember, this is Elon Musk we’re talking about.
The thing is that Grok-2 doesn’t have any filters. So, where other models would reject queries that are violent or political, Grok-2 doesn’t. While we’ve seen some obviously fake images like Mickey Mouse with a MAGA hat on smoking a cigarette, we all worry about the possibility of misinformation. This is a major election year, so the possibility of misinformation is at a height. This was true even before Grok-2 released.
Now that there’s a tool with no guardrails on the market we only expect more misguiding images to flood the internet. We’re not sure if any sort of regulatory institution is going to crack down on Elon’s model. We’ll have to wait for more information about this.
So far, along with the aforementioned images, people have been able to generate images of Donald Trump in a plane toward the Twin Towers, Bill Gates doing a line of cocaine, Pikachu in front of a Nintendo Store with a gun, and Kamala Harris in a boxing match with Donald Trump. It’s pretty easy to spot that these are fake because the prompts are nonsensical. Also, the image quality of the images lacks the refinement that makes tools like DALL-E and Gemini more of a threat.
The post Surprising no one, Grok generates reckless images appeared first on Android Headlines.