Detect Deepfakesby Resemble AI
Deepfake case study · Multi-modal

Update on Grok Imagine Filters (May 2026) — What’s Going On With the Colossus Migration

In January 2024, xAI tightened Grok's image generation filters after a deepfake controversy, restricting image generation, especially involving real people, due to legal reasons.

Incident date
Jan 2024
Target
real people
Updated May 23, 2026 · 1 min read

In early 2024, xAI's Grok faced scrutiny for generating sexualized deepfakes of real people. This led to tightened filters and restrictions on the AI's image generation capabilities.

What happened

Following a controversy involving deepfakes of real people, xAI significantly tightened the filters on Grok Imagine. This resulted in stricter limitations on image generation, especially concerning real individuals and content involving undressing or non-consensual scenarios. The restrictions were implemented in January. Later, in May, further changes were observed, seemingly related to xAI's migration from Colossus 1 to Colossus 2, which may temporarily reduce heavy processes to keep things stable, and that appears to be affecting image and video generation. Content involving real people is expected to remain heavily restricted due to legal concerns.

Sources