AI chatbots and image generators have become major players on the internet. X AI chatbot Grok is the latest to make headlines, with its new image-generating capabilities having stirred up nothing but chaos.
To access the bot, users have to be subscribed to X Premium, better known as the X (formerly Twitter) blue tick program. Grok has enabled X Premium users to generate seemingly countless images sans filters.
Images of Elon Musk supporting transgender community rights (which he notoriously refuses to do), Donald Trump holding Kamala Harris at gunpoint, and Barack Obama doing drugs have been posted freely on the platform.
This puts Elon Musk’s X under further scrutiny. With the 2024 election looming and European regulators already monitoring it closely, it looks like yet another battle for the social media platform is likely.
What are the limits of X’s AI image generator?
For users who have been puzzled about what Grok’s limits for image generation are, there is some user guidance.
First off, Grok avoids generating explicit, extremely violent, or hateful imagery. The AI tool will also not generate images that might engage in copyright violations.
This second part, however, has proven to be challenging for Grok, given that users have generated images of beloved characters doing questionable things.
An example is this picture of Mario, holding a beer and smoking a cigar on the beach. The image has attained over 1.3 million views and more than 27,000 likes.
Lastly, Grok also says it won’t generate images that may be used to deceive others, such as deep fakes.
X’s problem with prompt loopholes for image generation
For those unfamiliar with how image generation applications work, the user types a prompt for the image that he or she wishes to generate, and the AI will simply generate this. There are plenty of AI image generators, and they will not allow for the creation of certain sensitive content.
However, in the early days of apps such as Midjourney and DALLE-3, users could create loopholes to trick the AI into generating images that were otherwise off-limits.
Furthermore, X users have discovered that Grok can in fact quite easily be deceived into generating particular images. According to one user, if you tell Grok that you are a professional, the AI will allow users to generate just about anything.
It seems that many people are concerned about the state of the AI image generator, yet the owner of X, Elon Musk, is not. He has essentially said that this looseness has allowed people to “have some fun” with the generator.
