Tension between Apple and xAI: Grok escaped at the last minute

It has been revealed that Apple has taken drastic steps behind the scenes regarding the presence of the Grok chatbot developed by xAI in the App Store. According to information shared by NBC News, Apple seriously considered the possibility of removing the application from the store after the spread of sexualized deepfake content produced by Grok at the beginning of the year. Inappropriate edits, especially on photos of real people, were considered a violation of the company’s application policies.

The controversy, which started in the first months of the year when users asked Grok to nude the people in the photos, especially women and, in some cases, minors, quickly escalated. This development created public pressure on Apple to remove Grok and X applications from the App Store. Although Apple refrained from making a public statement, it appears that the company intervened in the background to solve the problem by directly contacting the developers.

According to NBC News, Apple contacted both the X and Grok teams and requested stricter measures regarding content moderation. The company asked developers to submit a concrete plan to address existing violations. In this process, the first update sent by X was not found sufficient and was rejected by Apple. Since the changes made did not meet expectations, new versions were prepared.

Additional regulations were made for Grok after Apple audits

According to the letter Apple sent to US senators, subsequent reviews found that the X app largely complied with the rules. However, the Grok application was not found sufficient at the same stage and it was stated that the violation of the App Store rules continued. Apple informed the developer and clearly stated that the application may be removed from the store if additional regulations are not made.

Following these warnings, new changes were made by Grok and the updated version of the application was re-evaluated. In the final stage, Apple decided that the improvements made were sufficient and approved the updated version of Grok. This process also clarified the background to the various moderation changes that xAI announced at that time. In this context, the limitation of access to visual production tools and the restrictions on the edits that can be made on photographs of real people were put into effect.

However, the current situation cannot be said to be completely problem-free. Grok can still produce sexualized images without users’ consent in some cases, according to a separate report from NBC News. Numerous examples documented over the last month show that existing restrictions can be bypassed by certain users. Although there has been a significant decrease in the number of such content compared to January, it is noteworthy that the system has not become completely secure.

The news states that some users may continue to produce images that show women in more revealing clothing. Edits made with clothing such as towels, sports bras or tight costumes reveal that the limits of existing filters can still be pushed. This situation once again reveals how complex content control in artificial intelligence-based visual production tools is and an area that needs to be constantly updated.