AI-Driven Cyber Deception and Exploitation Trends Emerge

• •GenPot uses LLMs to create convincing honeypots that deceive attackers.
• •Threat actors are embedding LLMs into their workflows for enhanced cyberattacks.
• •The use of AI in both defense and offense is reshaping cybersecurity strategies.

Researchers at the University of Málaga introduced GenPot, an AI-powered honeypot that deceived cybersecurity experts during trials, demonstrating the potential of large language models (LLMs) in cyber deception. Meanwhile, threat actors in Latin America are increasingly using commercial LLMs to automate post-exploitation processes, enhancing their attack capabilities. This shift indicates that LLMs are being leveraged not just for phishing but also for generating exploit scripts and facilitating data theft. The GenPot architecture separates transport, orchestration, and inference to improve safety and realism in honeypots, while attackers are embedding LLMs into their workflows to streamline operations. The impact of these developments is significant, with organizations needing to adapt to the evolving threat landscape where AI tools are being utilized by both defenders and attackers.