Musk's AI chatbot gets the reins
Elon Musk's AI chatbot «Grok» is being reined in after it initially generated shocking images of well-known figures. As a result, as of Friday, it was suddenly no longer possible to use text commands to have Grok generate images showing people using drugs or handling weapons.
Until then, there seemed to be hardly any restrictions in terms of content. Other offers for generating images from text specifications, such as Dall-E from ChatGPT inventor OpenAI or Imagen from Google, have guard rails that are designed to prevent this.
Musk, on the other hand, is known for his aversion to restrictions, which he describes as censorship. People should be able to have "a bit of fun" with the software, he wrote on Thursday. There have been no further comments from xAI so far.
Technology from German start-up
The actual technology used to generate the images does not come from xAI, but from the German start-up Black Forrest Labs. Musk's company relies on its "Flux.1" AI model. Musk wrote that his own software for generating images is being developed and will take a few more months.
The images generated by Grok sometimes have a photorealistic quality, but are often exaggerated like caricatures. Particularly ahead of the presidential election in the USA in November, there is great fear of AI forgeries that could influence public opinion. Many programs therefore avoid generating images with real people.
Companies fear for their reputation on X
The same applies to copyrighted characters such as Disney's Mickey Mouse, whose appearance in AI-generated images could have legal consequences. This could also increase the skepticism of major advertising customers towards the platform. Many of them already fear for the reputation of their brands and have withdrawn ads from X.
Musk recently sued several companies and an industry organization over this. He accuses them of a coordinated campaign to keep advertising dollars away from his platform. (SDA/swi)
