OpenAI, the organization responsible for ChatGPT, announced on Thursday that it has thwarted five covert influence operations in the past three months. These operations aimed to exploit OpenAI’s artificial intelligence models for deceptive purposes.
According to a blog post by OpenAI, the disrupted campaigns originated from Russia, China, Iran, and a private Israeli company. The threat actors attempted to utilize OpenAI’s powerful language models for tasks such as generating comments, articles, social media profiles, and debugging code for bots and websites.
CEO Sam Altman-led OpenAI stated that these operations did not seem to have significantly increased audience engagement or reach due to their services.
Concerns have been raised about companies like OpenAI amid fears that AI-powered tools could rapidly produce deceptive content in large volumes. This is particularly worrisome with major elections looming worldwide, as countries like Russia, China, and Iran are known for using covert social media campaigns to influence public opinion before voting.
Among the disrupted operations, one dubbed “Bad Grammar” was a previously undisclosed Russian campaign targeting various countries. It used OpenAI models to generate short political comments in Russian and English on platforms like Telegram.
Another well-known Russian operation, “Doppelganger,” utilized OpenAI’s AI to generate comments in multiple languages across platforms.
Additionally, OpenAI dismantled the Chinese “Spamouflage” operation, which exploited its models to conduct social media research and generate multi-language text.
An Iranian group, the “International Union of Virtual Media,” was also disrupted for using OpenAI to create content posted on Iranian state-linked websites.
Furthermore, OpenAI disrupted a commercial Israeli company called STOIC, which appeared to use its models to generate content across various social media platforms.
These operations were active on platforms like Twitter, Telegram, Facebook, and Medium, but none managed to engage a substantial audience, according to OpenAI.
The company highlighted AI leverage trends such as generating high volumes of text and images with fewer errors, mixing AI-generated content with traditional content, and faking engagement through AI replies.
OpenAI credited collaboration, intelligence sharing, and safeguards built into its models for enabling the disruptions.
