Open AI: “North Korean accounts that created fake resumes using ChatGPT have been removed.”
OpenAI deletes North Korean account that created fraudulent resumes using ChatGPT.
Open AI has revealed that it identified and removed North Korea-linked accounts that misused ChatGPT to create fake resumes. According to an official report released by Open AI on the 21st (local time), these North Korean accounts used ChatGPT to generate fraudulent resumes and online profiles, which were then posted on recruitment platforms in an attempt to illicitly secure jobs at Western companies. North Korea is known to leverage generative AI to steal advanced technology from the West or to secure funding for its illegal nuclear weapons program. However, Open AI did not disclose specific details regarding the number of North Korea-linked accounts removed or the timing of their deletion.
In addition to North Korean accounts, Open AI also disclosed the removal of accounts linked to China. Some Chinese accounts were found to have used ChatGPT to create Spanish-language news articles defaming the United States. These articles were published under the names of Chinese companies in major news outlets in South America. Furthermore, a Chinese security organization was discovered to have developed an AI-based surveillance tool using ChatGPT to monitor anti-China posts on social media. The tool was designed to collect and analyze posts and comments related to Chinese political and social themes on platforms such as X (formerly Twitter), Facebook, YouTube, and Instagram, with the intent of providing the data to Chinese authorities. Open AI referred to this practice as “Peer Review” and explained that it confirmed ChatGPT was used in the process of debugging (error correction) parts of the surveillance tool’s code.
Ben Nimmo, a senior researcher at Open AI, stated, “This is the first time an AI-based surveillance tool has been detected.” He added, “By tracking the traces left by malicious actors when they use AI models, we can understand their internet activities and detect harmful behavior.”