News, 03/05/2024, 12:00 PM
Between January and October of last year, more than 225,000 logs containing compromised credentials of OpenAI ChatGPT users appeared on dark web sites, Group-IB research revealed.
This data was found in logs associated with LummaC2, Raccoon and RedLine malware.
“The number of infected devices decreased slightly in the middle and late summer, but increased significantly between August and September,” the company said in the report.
Between June and October 2023, more than 130,000 devices with access to OpenAI ChatGPT were infected, a 36% increase compared to what was observed during the first five months of 2023. LummaC2 malware infected 70,484 devices, Raccoon malware 22,468, and RedLine 15,970.
“The sharp increase in the number of ChatGPT credentials for sale is due to the overall increase in the number of devices infected with information-stealing malware,” the Group-IB report states.
The data was released as Microsoft and OpenAI revealed that hackers from Russia, North Korea, Iran and China are experimenting with artificial intelligence (AI) and large-scale language models (LLM) to perfect their ongoing cyber attacks.
Group-IB says hackers can use LLMs to design convincing scams and phishing attacks and improve operational productivity.
“They are in the past [akteri pretnji] were mainly interested in corporate computers and systems with network access,” the report states. “Now they're also focusing on devices with access to public AI systems.”
“This gives them access to logs with a history of communication between employees and the system, which they can use to look for classified information (for espionage purposes), internal infrastructure details, authentication data (for conducting more dangerous attacks), and application source code information.” “
Photo: Matheus Bertelli / Pexels