[imagesource:x/@himbodhisattva]

Elon Musk’s foul-mouthed AI chatbot named Grok is letting anyone create images from text prompts and publish them to X, and so far, the rollout seems as rogue as everything else on the billionaire’s social network.

Subscribers to X Premium, which grants access to Grok, have been posting everything from Barack Obama holding a knife at Joe Biden’s neck to Donald Trump with a pregnant woman supposedly resembling Kamala Harris to Trump and Harris pointing guns as they fly towards the Twin Towers.

AKA, there don’t seem to be any restrictions to the kind of wild and fake images you can create and spread.

At least two images prompted by different users show Donald Trump caressing the belly of a pregnant Kamala Harris, which shows that this isn’t just a one-off case of the AI going off the rails.

it’s really good at this pic.twitter.com/gpAg54BWsg

— shb (@himbodhisattva) August 14, 2024

More disturbing is seeing the pair carrying out what appears to be the 9/11 attacks on the Twin Towers:

Grok will let you make anything right now pic.twitter.com/P2gALN2tzk

— Wynner (The Caretaker) 🇩🇪 (@EuroWynner) August 14, 2024

While these depictions are absurd, they are lifelike enough to demonstrate the AI’s potential to output misinformation, causing confusion and sewing deeper divisions. With US elections approaching and X already under scrutiny from regulators in Europe, it’s a recipe for a new fight over the risks of generative AI.

Musk’s ethos behind Grok, as it is behind his takeover of X, is that it’s meant to be the “anti-woke” and free-speech absolutist’s alternative to the censored chatbots offered by Silicon Valley competitors like OpenAI, per Futurism. This ultimately makes it quite vulgar, with flippant disregard for the rules, all with the potential to stir trouble.

Whereas most AI image generators, such as OpenAI’s DALL-E 3, prohibit images of public figures and illegal activities, Grok-2 is already churning out pictures of Trump dual-wielding revolvers and George W. Bush doing lines of coke.

Grok 2.0 …. Ohh boyyyy 😆😆😆 pic.twitter.com/TjzB7WMhVp

— Benjamin De Kraker 🏴‍☠️ (@BenjaminDEKR) August 14, 2024

Fans of Grok are electrified as these capabilities, free from the shackles of censorship, are a long-awaited dream come true, finally allowing them to unveil what the so-called “woke” mob would rather keep hidden. For some reason, this has to include an image of Trump and Obama making out.

It’s amazing to be able to generate images with Grok now, and not have censorship on the content and people! pic.twitter.com/i1VejnYIen

— John Faires (@johnfaires) August 14, 2024

Funny and silly, sure, but there is also far darker material being generated, like a “hyper-realistic” image of Adolf Hitler, and according to testing by The Verge, Trump in Nazi garb and more harrowing, Mickey Mouse gunning down children.

It appears as though Grok is now the only AI that will generate certain images deemed “unacceptable” by others!

Thanks @elonmusk!#Grok #AI #FreeSpeech pic.twitter.com/VWMrFOxrNy

— Crusader of Truth (@TruthXChrist) August 14, 2024

Even when Grok does refuse to generate something (with its terse rules, like avoiding generating images that are “pornographic, excessively violent, hateful, or that promote dangerous activities,” it says) loopholes are easy to find.

That leaves very few safeguards against it spitting out gory images of Musk and Mickey Mouse gunning down children, or even “child pornography if given the proper prompts,” according to X user Christian Montessori. And while Musk is aware of these issues, he seems to find them amusing, saying the tool is allowing people “to have some fun.”

Grok’s flimsy safeguards add fuel to the fire, pushing high-profile users and advertisers to distance themselves from X. Hopefully they can stay strong, even as Musk, undeterred, flexes his legal might, attempting to strong-arm them back into the fold.

[source:futurism]