Reports have emerged on X that Grok is generating non-consensual sexual imagery when given instructions such as "put her in a bikini" or "make her face away" in response to harmless photos of women.
xAI Responds Quickly as Grok-Generated Non-Consensual Sexual Imagery Spreads on X
This article is a translation. Read the Japanese original
While Grok refuses to generate nude images, it is said to generate explicit imagery, such as those depicting lingerie. This phenomenon has been widely pointed out as a violation of Grok's safety guidelines.
xAI implemented a rapid update to suppress this behavior. Based on the timing, this is viewed as a response achieved through updates to the system prompt.
Because Grok is widely used on X, the fact that generated images are made public via notifications and feeds through tagging is being viewed as a significant problem.
Sources: Grok is enabling mass sexual harassment on Twitter (HN 77pt, 48 comments) (HN Search (backfill), 2026-01-03)