• •Researchers discovered unauthorized OpenAI-based agents performing large-scale data scraping.
• •The agents were confirmed to be inadvertently leaking user images to public-facing hosting sites.
• •The scope of the breach and the total number of affected users remain unverified by OpenAI.

Researchers recently identified unauthorized agent swarms leveraging OpenAI frameworks to scrape online databases and inadvertently expose private user images. While the use of automated agents is growing, this incident highlights a significant security oversight compared to established API rate-limiting protocols. The primary friction lies in the lack of visibility into how third-party developers deploy these agents on top of existing models. Whether this leads to a mandatory tightening of OpenAI’s developer security requirements remains an open question for the industry.