From obviously AI-generated to photographically realistic, Meta’s new tool Muse Image can do both. Image: screenshot meta
With just a few clicks, users of Meta’s social media platforms will be able to generate deceptively real-looking images of Meta users who have not set their account to private.
July 12, 2026, 6:01 p.mJuly 12, 2026, 6:01 p.m
We already encounter AI-generated images and videos in many places in our everyday lives. The dimensions range from village festival flyers to cat videos to national advertising campaigns from SBB or Swisscom.
While AI-generated content in advertising is primarily criticized because it replaces creative work and deprives artists of work, the use of so-called text-to-image tools in the private sector poses additional risks of misuse of the technology.
The ability to use text input to create deceptively realistic images of real people in situations they have never found themselves in increases the risk of identity abuse, the spread of misinformation, and the production and distribution of AI-generated, non-consensual pornographic content.
Elon Musk’s Grok is followed by Meta’s Muse Image
Just recently did a report in the US magazine “The Information” Headlines in which two former employees of xAI, Elon Musk’s AI company, reported that around half of all requests for image generation were pornographic in nature. Grok is integrated directly into Musk’s social media platform X, so problematic fictional content can be created seamlessly and shared immediately.
In Times Square, activists protest against his AI tool Grok with an inflatable Musk doll.Image: imago-images.de
Now, Meta, the next tech giant has released its own text-to-image tool and integrated it directly into the company’s own social platforms and messenger services. The tool called Muse Image can currently be used via the web browser and the Meta-AI app. In the USA, the function is already embedded in the story function on WhatsApp and Instagram.
In a blog post, Mark Zuckerberg’s tech company writes: “Muse Image acts as a creative partner that knows your world and makes it easy for you to transform your ideas into high-quality graphics that you can download and share anywhere, for example directly in your feed, story or chat.” The company has also announced that it will soon offer the function in other countries, expand the functions to Instagram and also offer the tool for Facebook and Messenger. A version for video generation is already in the works.
AI tool uses user profile pictures
After Meta introduced its new software, one feature in particular caused a stir. Users can mark other users in their text input with an “@” and the corresponding profile name and the AI immediately creates a picture of the marked user using the person’s profile picture. In order to prevent their own images from being used for such requests, users must either set their account to private or actively prohibit the use of their own images in the settings.
Pictures of other public users can be easily created by specifying the profile name. Image: screenshot meta
Image generation tools like Metas Muse Image are nothing new. As early as 2022, the first powerful text-to-image tools came onto the market with DALL-E 2 or Midjourney, which enabled increasingly realistic image generation for a wide range of people. The integration of generative image AI into social networks initially began with relatively simple creative functions such as stylized avatars, filters or personalized image content.
With the integration of Grok into X in 2024, the focus shifted more towards creating photorealistic images directly within social networks. Due to the comparatively few restrictions on image generation, Grok has repeatedly come under criticism since its introduction because of the deepfakes and non-consensual sexualized content it creates. It was only this year that the EU initiated ongoing proceedings against platform operator X, and Malaysia and Indonesia temporarily restricted access to Grok entirely.
Data protection advocates warn about new AI tool
Although it is to be expected that Meta’s new Muse image function will be more heavily regulated in this regard than Elon Musk’s Grok, data protection advocates are looking at Meta’s latest development with concern. This also applies to Martin Steiger, lawyer for law in the digital space and spokesman for the data protection organization Digitale Gesellschaft Schweiz.
The lawyer criticizes the increased integration of such services on social media platforms: “As a result, the tools are suddenly directly available to billions of people, and the inhibition threshold for using them is completely eliminated.”
Martin Steiger is a lawyer and spokesman for the Swiss Digital Society. Image: KEYSTONE
The tools are also designed for high virality, and users are encouraged to share and distribute the images created directly, including problematic content: “Since AI is trained with large amounts of data that also contain social prejudices, this sometimes results in clichéd, racist or sexist representations.”
Markus Steiger sharply criticizes the fact that social media users who have a public account have to take action themselves to limit the powers of tech companies:
“With opt-out, Meta and other tech companies are relying on the inertia of the masses. Very few users change the often hidden default settings, while new AI functions are constantly activated automatically. Even critical users give up at some point.”
According to lawyer Martin Steiger, the big tech companies would simply ignore the legally anchored principle of “privacy by default”, i.e. that the standard settings must ensure data protection as well as possible.
You can deactivate the function here:
On Instagram, users with public accounts can prohibit the use of their own images under Settings and the Share and Reuse subcategory Muse Image. Image: screenshot instagram
In an article from the English news channel BBC The director of the data protection organization Foxglove, Donald Campbell, calls the new tool a “recipe for disaster”. With regard to the various incidents already known regarding non-consensual, AI-altered images being shared on social media, Campbell told the BBC: “It’s hard to understand why Mark Zuckerberg thinks it’s a good idea to further promote this creepy image manipulation.”