Meta is suing the maker of the controversial “nudify” app Crush AI, accusing it of distributing thousands of ads on its platforms. The lawsuit was filed in Hong Kong and targets Joy Timeline HK , the company allegedly behind the app.
According to Meta, Joy Timeline HK systematically circumvented the tech giant's ad controls by promoting services that use Generative AI to produce fake, sexually explicit images of real people, without their consent.
Although Meta claims to have repeatedly taken down ads that violated its policies, Joy Timeline HK allegedly continued to run new campaigns, bypassing the company's filters and security mechanisms.
See also: Arrests for trafficking in child sexual abuse material
Crush AI is said to have run over 8,000 ads for its “AI undresser” services in the first two weeks of 2025. According to Faked Up newsletter writer Alexios Mantzarlis, around 90% of Crush AI’s website traffic came from Facebook and Instagram.
Crush AI managed to advertise its services by creating dozens of new advertising accounts and constantly changing domain names. Some of the accounts reportedly had names like “Eraser Anyone's Clothes” accompanied by different numbers, while Crush AI also maintained a Facebook page to promote its service, boosting its visibility and reach to unsuspecting users.
Meta announced that it is considering new measures to strengthen security against applications that abuse artificial intelligence to compromise personal data or spread harmful content.
The phenomenon, however, is not limited to the Meta ecosystem. As platforms like X (formerly Twitter), Reddit , and YouTube try to integrate Generative AI features, they are facing a growing wave of AI abuse to create harmful content, with particular concern for the safety of minors. Researchers found that links to AI nudify apps spiked in 2024 on the above platforms, and millions of people reportedly received ads for such apps.
See also: Spain: Minors convicted for creating nude photos of girls using AI
In response to this growing problem, Meta and TikTok have banned keyword searches for AI nudify apps, but completely removing these services from their platforms has proven difficult.
In a recent post, Meta announced that it has developed new detection algorithmscapable of identifying ads for such apps, even when they do not contain explicit content. At the same time, the company has expanded the list of terms, phrases and emoji that trigger its filtering mechanisms, as well as a matching technology that accelerates the removal of identical ads from the platform. As the company revealed, since early 2025 it has identified and dismantled four separate networks of accounts that were distributing ads for such services.
Collaboration with other companies
In addition to the actions on its platforms, Meta announced that it is starting to share data on suspected nudify apps through Lantern, a joint effort by the Tech Coalition that includes tech giants like Google, Snap and others. The initiative aims to prevent online sexual exploitation of minors. As of March, Meta has already contributed more than 3,800 unique URLs to the program’s database.
Meta's moves are positive and necessary, but they come at a time when the race is already well underway. Social media platforms have proven to be extremely vulnerable to privacy-invading artificial intelligence, with nudify apps like Crush AI being a prime example.
See also: A new iMessage feature encourages kids to report inappropriate images
The fact that Meta is beefing up security, removing ads, and expanding its partnership with Lantern shows that it recognizes the seriousness of the phenomenon. However, the success of these measures will depend on two key factors:
- The speed and accuracy of detection systems — malicious AI tool creators are constantly changing tactics, and so should platforms.
- Transparency and cooperation between companies and states — without common standards and data sharing, the effort will be fragmented.
Overall, this is a step in the right direction, but more as a response to an already existing and serious problem, rather than as a preventive policy. Artificial intelligence requires upfront regulation, not just repression after the damage has been done.
Source: techcrunch.com
