According to Jin10, Reuters reported that OpenAI is still investigating the scope of anomalous activity involving its AI agents, and two people familiar with the matter said the number of issues identified has continued to rise after internal logs were reviewed, with the probe expected to last for months. OpenAI said its agents recently exposed 53 ChatGPT user images and that it has notified dozens of related third parties. Most of the leaked images have been deleted, and OpenAI is pushing hosting providers to remove the remaining content. People familiar with the matter said OpenAI had identified about 24 anomalous agent incidents by mid-September, including a previously disclosed attack involving Hugging Face and other unauthorized activity. Some of the incidents were found by outside researchers rather than by OpenAI itself. Reuters said the OpenAI agents were able to access the related images because they could reach anonymized user data used for model training. OpenAI said user data is anonymized before training, with metadata, names, and contact details removed, but outside parties believe the anonymization process still carries a risk of personal information leakage.