Grok accounts for 87% of deepfake attack files in H1 2026

A report by Resemble AI reveals that Elon Musk’s AI tool, Grok, accounted for 87% of AI-generated files linked to deepfake attacks in the first half of 2026. Researchers identified 821 independent deepfake incidents involving at least 15,736 victims, comprising approximately 3.46 million synthetic images, videos, and audio files.

Of the identified attacks, 137 cases involved non-consensual adult or child sexual imagery, representing about one-sixth of the total. Grok has previously faced criticism and regulatory scrutiny regarding users attempting to generate digital nudity or pornographic content. While direct financial losses from these attacks totaled $6.95 million, the primary harms identified were reputation damage, harassment, and sexual exploitation. Resemble AI estimates that companies allowing or spreading such imagery could face potential civil liabilities as high as $2.24 billion.

The report also notes a massive expansion in the reach of deepfake content, with potential exposure reaching 292.8 billion people in the first half of the year. Conversely, publicly reported corporate deepfake fraud has significantly declined, with only 15 reported cases, suggesting such incidents may be occurring less frequently in the public eye due to corporate reluctance to disclose attacks to protect reputations and investor confidence.