Apple almost dropped Grok from the App Store amid deepfake fury

Apple may have appeared restrained during the uproar over Grok’s deepfake controversy earlier this year, but newly revealed details suggest the company was far more active behind the scenes than it publicly indicated.

According to a letter obtained by NBC News, Apple identified violations of its App Store policies and warned that Grok could be removed if corrective steps were not taken. The intervention came after a wave of complaints and media reports highlighted how the chatbot could generate sexualised images of individuals, often without consent.

The issue quickly escalated globally into a broader debate about AI safety, particularly as users demonstrated how easily the tool could be used to manipulate photos, including those involving women and, in some instances, minors.

Behind-the-scenes pressure on Grok and X

While Apple did not initially comment publicly, it reportedly engaged directly with the teams behind both Grok and its parent platform, X, after the controversy gained traction. The company required developers to outline a clear plan to address content moderation failures.

More from Tech

Subsequent app updates submitted for review did not immediately meet Apple’s standards. One version was rejected outright, with the company determining that the proposed fixes were insufficient to bring the app into compliance with its guidelines.

Only after further revisions did Apple approve a new version of the Grok app, concluding that the changes had substantially improved its moderation systems.

However, the process revealed how close the app came to being removed entirely from the App Store.

These newly disclosed interactions help explain the rapid and, at times, confusing adjustments made by xAI during the height of the backlash. Restrictions were introduced on who could access certain image-generation features, alongside tighter controls on editing real people’s photos.

Deepfake concerns persist despite safeguards

Despite these efforts, concerns around Grok’s image-generation capabilities have not fully subsided. Firstpost found that Grok was still producing such sexualised images without consent in certain cases, even after moderation updates were rolled out.

Although the overall volume of such content has reportedly declined since the peak of the controversy, some users have still been able to bypass safeguards. In particular, prompts that subtly alter clothing or appearance have allowed the generation of suggestive images featuring real individuals.

This persistence highlights the ongoing challenge faced by AI developers and platform operators alike. As generative tools become more sophisticated, enforcing consistent and effective moderation remains a complex task.

For Apple, the episode underscores the growing responsibility of app marketplaces in policing AI-driven content. For developers, it serves as a reminder that compliance is not a one-time fix but an ongoing process, especially as users continuously test the limits of emerging technologies.