Safety researchers report that leading AI image generators from OpenAI and Google are becoming more susceptible to producing explicit content, even as these models grow more capable. This vulnerability persists despite ongoing warnings about the unchecked spread of AI-generated explicit material. A study co-authored by Google, published in July, examined internet users across the United States, Australia, and the United Kingdom. The research found that much of this non-consensual content is shared through text and direct messages or stored in private clouds, making it less likely to be detected or reported. Consequently, Google researchers concluded that the most critical intervention point remains at the source, specifically the harmful AI tools used to create fake sexual images. Recent red-teaming efforts suggest that while the image generation capabilities of these leading models have improved, their ability to block explicit outputs has declined. Regulators have not yet taken meaningful steps to stop this trend, leaving the problem largely unaddressed.