Hackers Deploy AI-Powered Malware That Writes Code in Real-Time as State Actors Exploit ChatGPT for Cyberattacks

Hackers unleash revolutionary AI-powered malware that writes malicious code in real-time while state actors from North Korea, Iran, and China exploit ChatGPT and other AI tools through social engineering tactics, creating a dangerous new era of dynamic cyberattacks that adapt and evolve during execution.

• Google Threat Intelligence Group identifies threat actors deploying novel AI-enabled malware like PROMPTFLUX and PROMPTSTEAL that use Large Language Models during execution to dynamically generate malicious scripts and obfuscate code in real-time
• Cybercriminals are using social engineering tactics to bypass AI safety guardrails by posing as students in cybersecurity competitions or researchers to trick Gemini into providing information for vulnerability exploitation and tool development
• State-sponsored actors from North Korea, Iran, and China continue misusing AI tools across entire attack lifecycles while underground marketplaces for illicit AI tools mature with multifunctional offerings supporting phishing, malware development, and vulnerability research