Elon Musk’s AI Image Generator Will Make Pictures of Almost Anything
With concerns over AI being misused, most companies are putting guardrails in place to curb the technology’s worst excesses. Not so at Elon Musk’s X (formerly Twitter).
X launched an updated version of its AI model Grok this week that comes with new image generation capabilities powered by FLUX.1.
Users quickly discovered there are little to no safeguards on the AI image generator. Some began making photos of a pregnant Kamala Harris with Donald Trump, Mickey Mouse on a killing spree, and Barack Obama doing cocaine.
Most AI image generators have protections in place to prevent users from making pictures of trademarked characters, real-life people, or illegal activities. OpenAI, for example, denies requests for images of a particular person or entity on its image generator DALL-E.
But Musk’s X is taking a different approach. The platform is being flooded with wild images of beloved characters like Pikachu and Goofy doing unspeakable things.
Oh my god. Grok has absolutely no filters for its image generation. This is one of the most reckless and irresponsible AI implementations I've ever seen. pic.twitter.com/oiyRhW5jpF
— Alejandra Caraballo (@Esqueer_) August 14, 2024
The image generations are also of excellent quality. PetaPixel reported last week that the generative AI model behind Grok, FLUX.1, is a new AI image generator from the same researchers who developed Stable Diffusion and invented the latent diffusion technique.
FlUX.1 has almost certainly been trained on an enormous amount of copyrighted images scraped from the internet.
Grok is exclusively available to subscribers of X Premium (which costs $8 per month). The Verge has access to it and the tech publication was able to generate an image of “Donald Trump wearing a Nazi uniform”, “Antifa curbstomping a police officer”, and “Bill Gates sniffing a line of cocaine from a table with a Microsoft logo”.
The Verge also generated an image of “sexy Taylor Swift” which resulted in the pop star wearing a semi-transparent black lace bra. At the beginning of the year, there was an uproar when explicit, deepfake images of Swift were shared on X. However, Grok apparently stops when asked to generate nudity.
How long Musk’s gung-ho approach to AI images is anyone’s guess. X is currently under investigation by the European Commission for potential violations of the Digital Safety Act. While in the U.S., legislators are seeking ways to regulate deepfakes and AI-generated misinformation.
