ALBAWABA – OpenAI has said in a blogpost that in collaboration with Microsoft, they were able to identify and block five threat groups backed by nations that have been found to be using their Large Language Models in malicious activities such as preparing Phishing Scams, translating technical papers and writing and debugging potentially harmful codes.
Through cooperation and information exchange with Microsoft, OpenAI says they were able to take down five malicious actors with ties to states: two threat actors from China, identified as Charcoal Typhoon and Salmon Typhoon; Crimson Sandstorm from Iran; Emerald Sleet from North Korea; and Forest Blizzard from Russia, stressing that the OpenAI accounts linked to these perpetrators were closed after identification.
Microsoft notes in a separate blogpost that “Over the last year, the speed, scale, and sophistication of attacks has increased alongside the rapid development and adoption of AI,” adding that Microsoft and OpenAI have not yet noticed any especially unusual, cutting-edge attack or misuse methods made possible by threat actors' use of AI.
Liu Pengyu, a spokesman for the Chinese embassy in the United States, according to Reuters, stated that China rejects "groundless smears and accusations against China" and supports the "safe, reliable, and controllable" application of AI technology to "enhance the common well-being of all mankind."
Tom Burt, Customer Security Vice President at Microsoft, spoke to Reuters about the incident, stating that none of the threat actors were able to make any advanced leaps, saying “we really saw them just using this technology like any other user, however Microsoft points out that they track more than 160 nation-state threat actors, with OpenAI noting that when these groups’ activity is detected, the company intervenes appropriately by blocking their accounts, stopping services, or restricting their access to resources.
