Kaspersky's Digital Footprint Intelligence service has identified nearly 3,000 posts on the dark web that mainly discuss the use of ChatGPT and other LLMs for illegal activities. Perpetrators are looking for ways to deceive, from creating malicious alternatives to the chatbot to jailbreaking the original and beyond. Stolen ChatGPT accounts and services offering their mass automated creation are also flooding the dark web channels, reaching an additional 3,000 posts.
In 2023, Kaspersky Digital Footprint Intelligence discovered nearly 3,000 posts on the dark web discussing the use of ChatGPT for illegal purposes or talking about AI-based tools. Although the conversations peaked in March, the discussions are still going on.
In addition to the chatbots and AI mentioned, significant attention is being paid to projects such as XXXGPT, FraudGPT and others.These language models are being promoted on the dark web as alternatives to ChatGPT, having additional features and without the original limitations.
Another threat to users and companies is the purchase of accounts for the paid version of ChatGPT. In 2023, another 3,000 posts (in addition to the ones mentioned above) were identified advertising ChatGPT accounts for sale across the dark web and on covert Telegram channels. These posts either distribute stolen accounts or promote auto-enrollment services that create accounts en masse upon request. It is worth noting that specific posts have been republished on many dark web channels.
To avoid the threats related to the activities of cybercriminals in the dangerous part of the internet, you can implement the following security measures: