Microsoft and OpenAI have identified five state-sponsored groups that are utilizing generative AI, potentially including ChatGPT, to enhance their hacking capabilities. The groups consist of well-known Russian hacking group Fancy Bear, as well as hackers from China, Iran, and North Korea, according to research published by the companies on Wednesday.
These hackers attempted to leverage OpenAI’s tools for tasks such as debugging computer code, generating computer programs, and creating content, likely for phishing attacks. Fancy Bear hackers even employed the tools for open-source research into satellite communication protocols and radar imaging technology.
Meanwhile, the North Korean hacking group Emerald Sleet utilized AI technology to identify experts and organizations focused on defense in the Asia-Pacific region.
Although this news may raise concerns, OpenAI reassures that their models offer only limited capabilities for malicious cybersecurity tasks. Microsoft also emphasizes that the AI tools did not result in any particularly novel or unique AI-enabled attacks. Instead, the hackers were simply using generative AI as a productivity tool to streamline their work.
In response, both Microsoft and OpenAI have terminated the accounts used by the hackers to access generative AI tools. They are also actively monitoring their services to prevent any potential misuse.
This news comes at a time when intelligence services in the UK and South Korea have also detected state-sponsored hackers utilizing generative AI. The worry is that tools like ChatGPT can rapidly create content, analyze databases, and identify software bugs, making them powerful weapons for cybercriminals.
However, OpenAI assures that it has implemented safeguards across its tools to prevent malicious misuse. Their latest AI model, GPT-4, offers only limited capabilities for malicious cybersecurity tasks beyond what is already achievable with publicly available, non-AI powered tools.
OpenAI and Microsoft have collaborated on this research to help the industry stay ahead of potential misuse of AI tools. Redmond adds that they will collaborate with other stakeholders to regularly exchange information about detected threat actors’ use of AI.