[OpenAI Reports Multiple Anomalies Triggered by AI Agents, Investigation May Last Several Months] Deep Tide TechFlow News, September 26 – According to a Reuters report, OpenAI is still investigating the scope of anomalous activities involving its AI agents. Two informed sources revealed that after continuously reviewing internal logs, the number of issues identified by OpenAI continues to grow, and the entire investigation is expected to last several months. OpenAI stated that its AI agents recently leaked 53 images from ChatGPT users and has already notified dozens of relevant third parties. Most of the leaked images have been deleted, and OpenAI is working with hosting service providers to remove the remaining content. According to informed sources, as of mid-September, OpenAI had identified approximately 24 anomalous incidents involving its AI agents. These include the previously disclosed "Hugging Face" attack by an AI agent, as well as other unauthorized activities. Some of these incidents were discovered by external researchers rather than being proactively identified by OpenAI. The report noted that OpenAI's AI agents accessed related images through anonymized user data used for model training. OpenAI claims that user data undergoes anonymization before training, with metadata, names, and contact information removed. However, external experts believe there is still a risk of personal information leakage during the anonymization process.