Law enforcement agencies and cybersecurity researchers are raising the alarm about a disturbing trend: cybercriminals are increasingly using generative AI tools to write functional, customized malware, significantly lowering the technical barrier to entry for would-be attackers. Cases have been documented in which individuals with no prior programming experience have used AI systems to create working malicious software.

While major AI providers have implemented safety guardrails designed to prevent their tools from generating harmful code, researchers have demonstrated multiple techniques for bypassing these restrictions, including prompt injection attacks, character substitution tricks, and the use of jailbroken model variants circulating in criminal communities.

"Generative AI is a force multiplier for threat actors," said cybercrime investigator Detective Sergeant Kyle Morrison. "We're seeing malware that shows signs of AI assistance in its development — cleaner code structure, better obfuscation, more sophisticated evasion techniques — from individuals who clearly don't have the traditional hacking background."

The malware produced with AI assistance ranges from simple keyloggers and ransomware variants to more complex tools designed to evade specific security products. In some cases, the AI has been used to write malware in less common programming languages that are less well-detected by antivirus software.

AI developers are under increasing pressure from security researchers and policymakers to strengthen safeguards against misuse. Law enforcement agencies are also calling for better mechanisms to identify and track AI-assisted malicious software in the wild.