xthe old Twittersurprised its free account users this Monday by giving them access to generative AI Grokso far only available to X Premium subscribers. Grok, currently in its version 2, has arrived accompanied by a new model for generating images from text instructions called Aurorawhich is new for both free and paid account users. Aurora is attracting a lot of attention for several reasons: its accessibility, its quality close to photorealism at times and the permissiveness of the platform when generating controversial content.
It is well known that X is a platform with fewer restrictions than others when publishing content, an approach that carries over to the field of AI image creation. While with ChatGPT or Gemini trying to generate an image with a celebrity will not work, Aurora has no problem. Historical figures, film, television and music stars, athletes or politicians are represented by Grok, often very convincingly, without problems.
no, these are not photographs. this is Grok 2 + Aurora pic.twitter.com/mgjRumCbex
— ale𝕏 fazio (@alxfazio) December 7, 2024
Very impressive Aurora, Grok’s hyper-realistic function to generate AI images with familiar faces pic.twitter.com/cdq9eYAVuq
— Ricardo Sametband (@rsametband) December 10, 2024
Logically, there are some limits. It is not possible to create nudity, scenes of direct violence or drug consumption, but it is possible to approach the limit, suggest and achieve a similar result without breaking the usage policies.. For example, if you want to show a well-known character snorting cocaine, Grok will not represent the action but will show the subject in front of the line of cocaine, which means more or less the same thing.
With these wickers, as expected, social networks are filling up with all kinds of images with famous personalities of all kinds and in bizarre situations. In Spainwhere Twitter AI has been trending all week, Pedro Sanchez and Franc They are the protagonists of many of the images in a variety of situations, probably motivated by the announcement that in 2025 the Government will dedicate a lot of time, effort and taxpayer money to commemorate Franco’s death half a century ago.
testing #Grok ➡️Tribute for the 50th anniversary of Franco’s death. pic.twitter.com/Q1CDwo1VEr
— Almirón (@vic_almiron) December 11, 2024
What would Franco say to Sánchez?
I start: “Pedro, thank you for dedicating more time to me than to the DANA victims.” pic.twitter.com/qYHMg0Bbfg
— Muy.Mona/🇪🇸💚 (@Capitana_espana) December 11, 2024
Other dictators, such as Stalin and hitlerare also common in the images generated with Grok that go viral. Just as those of historical footballers that did not coincide in time or in teams in which they have never been. Images of famous historical figures, politicians, actresses and singerss dressed, again, in equipment from different football clubs.
Twitter’s AI is very tough, some examples: pic.twitter.com/obX7dqLONP
— Ilicitano Museum (@MuseoElche) December 11, 2024
God, Twitter’s AI is spectacular HAHAHAHAHAHAHAHAHAHA pic.twitter.com/yrdVImfPfH
— 🥊𝕳 BARRAS DEL MUNDO ⚽🍺 (@Barras_LATAM) December 10, 2024
I love you so much Twitter AI pic.twitter.com/A3aPWc7kEb
— Mr_Pacharanes (@GxldeCerezo) December 10, 2024
Let’s make a thread of famous people with the America shirt using Twitter’s AI Grok, I’ll start🧵
Billie Eilish💛 pic.twitter.com/z9X4IY6inC
— Roberto Haz (@tudimebeto) December 11, 2024
The greatest prominence goes to the best-known personalities: Vladimir Putin, Donald Trump, Taylor Swift, Joe Biden, Elon Musk, Diego Maradona, Cristiano Ronaldo, Lionel Messi, Will Smith and many others.
Twitter’s AI is going to drive journalists around the world crazy…
Yes, in its first versions it is already capable of doing this.
Very soon we will not be able to know what is real and what is not. pic.twitter.com/gP9hMIBSsM
— Germán Huertas (@Nocodeboy) December 12, 2024
Since generative artificial intelligence broke into the technological mainstream 2 years ago, there has been warning of the risk they pose due to their use to produce misinformation content. Grok 2 will make many shake their heads and it would not be strange if the EU ends up taking measures against X for this AI: for training it with the content of the platform’s users by default, only with the possibility of avoiding it from the application settings, and for the platform’s contribution to the always elastic and adaptable term of misinformation. Meanwhile, there will be no shortage of fun.