shelLM is an AI-based honeypot created to engage attackers as a real system. It was developed by cybersecurity researchers from various universities and organizations. The researchers aimed to make the honeypot indistinguishable from a real system by using prompts, detailed behavior descriptions, and a Chain of Thought approach. The honeypot was tested with 12 users, and they found that it had a true negative rate of 90% and a false positive rate of 9%. The top ten commands used in the honeypot were cat, ls, sudo, get, echo, pwd, nano, ping, ssh, and whois. The researchers achieved a 92% accuracy in generating synthetic data for the honeypot. #AI #Cybersecurity #honeypot #shelLM

https://cybersecuritynews.com/shellm-ai-based-honeypot-engage-attackers/