Detect Deepfakesby Resemble AI
Deepfake case study · Multi-modal

Grok continue d'héberger des deepfakes sexualisés de femmes célèbres

An investigation reveals that xAI tool Grok Imagine continues to generate and host non-consensual sexualized deepfakes of famous women and political figures

Incident date
Jun 2026
Target
famous women and political figures
Updated Jun 12, 2026 · 1 min read

On June 12, 2026, reports highlighted that the xAI tool Grok Imagine persists in generating and hosting explicit deepfake images and videos of celebrities and political figures without their consent. Despite previous commitments to implement security guardrails, the platform remains a source of non-consensual sexualized content.

What happened

Although some users have noted increased moderation efforts, experts and researchers argue that the current safety measures remain insufficient compared to broader industry standards. The investigation found that users are successfully bypassing existing filters by employing diverted prompts to generate prohibited imagery. This ongoing vulnerability has drawn sharp criticism from digital rights organizations and international regulators, including those in Canada. The potential for legal and reputational damage has reached a significant scale; notably, SpaceX has allocated a financial reserve of $530 million to address litigation risks related to the use of Grok for creating harmful content.

Sources