Grok, the artificial intelligence (AI) from xAI, is now available free to all users of X, formerly Twitter. It also includes an image generator which, obviously, does not have many safeguards.
Risks of misinformation
Indeed, Grok allows you to generate ultra-realistic images of personalities. They range from international celebrities to local figures. And France is obviously not immune to the phenomenon; since AI has been available for free, false images of people known within France abound on the platform.
(THREAD)
Like many, I test Grok and observe his evolution day by day. A few months ago, with the exception of high-profile figures like Trump or Musk, AI had difficulty generating lesser-known faces.
It’s getting better and it’s funny, so I’m sharing. pic.twitter.com/wU4voFBJlH
— The Toad (@Zoltan_47) December 11, 2024
Unsurprisingly, political figures are very popular. Eric Ciotti as a Jedi, Jean-Luc Mélenchon and Marine Le Pen hugging, Pascal Praud and Mathilde Panot sharing a drink… Internet users clearly don’t lack imagination. Actors and TV presenters are also portrayed.
Sure, this feature looks entertaining on the surface, but it’s also a testament to Grok’s power, with some of the visuals looking really close to reality. The risks for the spread of disinformation are gigantic. “ Those who are familiar with AI image generators can spot with enough confidence whether the image is fake or not. But I worry about us mere mortals who can’t », Remarks a user of the social network.
Because it is also possible to generate visuals of highly controversial personalitieslike Adolf Hitler, Osama Bin Laden or Stalin.
Very strong in replicating known faces, but really catastrophic in taking the environment into account realistically.
A sort of luxury photomontage, but which seems to “cheat” somewhere. pic.twitter.com/WgAkOt2HAc
— Raphael Grably (@GrablyR) December 12, 2024
Accessible to millions of people
This completely deregulated approach corresponds to the idea defended by Elon Musk since the creation of xAI. The startup aims to compete with OpenAI, which it considers too “ politically correct “. Moreover, the billionaire says he is a fervent defender of absolute freedom of expression, hence the almost non-existent moderation applied on the former Twitter.
The image generators of its rivals, like Google or Meta, are much more controlled and equipped with numerous filters, preventing users from creating visuals if they type certain keywords. With Grok, it’s very visibly the opposite. And the fact that AI is accessible to so many people is very worrying. For comparison, Midjourney, which also allows you to generate images of public figures, is paid.
From now on, all X users, that is to say several million people around the world, can use the image generator as they wish. It is very likely that some will not necessarily have benevolent ideas in mind.
Grok, which is trained on data present on the social network, should improve drastically. xAI now leverages the world’s most powerful supercomputer to train its artificial intelligence.
- Grok, Elon Musk’s artificial intelligence, is now available for free to all X users.
- It allows you to generate images of well-known personalities, even in France.
- The risks of misinformation are immense, especially since the visuals are ultra realistic.
