On Wednesday, OpenAI announced that since the year's beginning, it has thwarted over 20 operations and fraudulent networks that sought to utilize its platform for malevolent ends worldwide.

This work involved creating bios for social media profiles, diagnosing malware, creating content for websites, and creating artificial intelligence (AI) generated profile pictures for fake accounts on X.

According to the artificial intelligence (AI) company, threat actors are still experimenting and evolving with their models, but they haven't seen any indication that this is resulting in any significant advances in their capacity to produce much more malware or generate viral audiences.

It further stated that it obstructed the production of social media content concerning elections in the US, Rwanda, and, to a lesser degree, India and the EU, and that not a single network was able to attract repeat visitors or viral interaction.

Earlier this May, Meta and OpenAI revealed that an Israeli commercial company called STOIC (also known as Zero Zeno) was involved in efforts to produce social media comments about the Indian elections.

Here are a few of the cyber operations that OpenAI has identified:

• Artificial intelligence models were employed by SweetSpecter, an attacker who may have originated in China, for a number of goals, including vulnerability research, anomaly detection evasion, scripting support, and LLM-informed surveillance. It has also observed failed spear-phishing attempts made against OpenAI employees in an effort to spread the SugarGh0st RAT.
• Cyber Av3ngers, an organization affiliated with the Iranian Islamic Revolutionary Guard Corps (IRGC), used its artificial intelligence models to study programmable logic controllers.
• Storm-0817 is an Iranian threat actor that translated LinkedIn profiles into Persian and utilized artificial intelligence models to debug Android malware that could gather private data. It also developed tools to scrape Instagram accounts using Selenium.

A2Z and Stop News, two influence operations that produced content in both English and French for later distribution on numerous websites and social media accounts across multiple platforms, were among the clusters of accounts that the company said it had taken action to block.