Elon Musk's Grok dominates alarming deepfake report — 1 in 6 attacks involved sexual images including children

Elon Musk's AI-powered chatbot, Grok, generated 87% of deepfake attacks during the first half of 2026, according to a new report from Resemble AI. Researchers analyzed 1,760 news reports and discovered 821 separate incidents involving at least 15,736 victims, encompassing approximately 3.46 million synthetic images, videos, and audio files.

Even though the dataset is based on publicly reported cases, the percentage only applies to files Resemble AI could count and link to a specific tool. The findings underscore how rapidly AI-generated image generators can be exploited for harmful purposes. Among the 821 attacks, about one in six involved non-consensual sexual imagery of adults or children.

This triggered investigations by regulators worldwide, with the European Commission and California Attorney General launching formal inquiries. Companies potentially face up to $2.24 billion in civil liability under US law for distributing these images, although actual financial losses were smaller, amounting to $6.95 million. Many incidents led to reputational damage, harassment, or sexual exploitation rather than straightforward financial fraud.

Deepfake attacks are becoming more challenging to track, with Resemble AI estimating a potential media reach of 292.8 billion during the first half of the year, nearly matching the 296.4 billion from 2025. Political deepfakes were the most widely reported, with 22 incidents reaching over 1 billion people each. However, corporate deepfake fraud seems to have diminished in public view, with only 15 incidents reported, all from a single publication.

Resemble AI launched DETECT-World, a system designed to identify AI-generated content it hasn't seen before. It combines AI signatures with physical scene analysis, achieving 95.8% accuracy for images and 98.2% for video in internal tests.