AI-Written Malware Is Here — But Your Existing Defences May Already Stop It

Cybersecurity researchers at Unit 42 have published new findings on how attackers are using artificial intelligence to write malicious code, a trend they describe as evolving 'from brand abuse to agentic execution.' The research highlights that AI tools are increasingly being used to help criminals create malware, adding a new layer of concern for businesses already managing cyber risk.

However, the report also offers some reassurance: existing behavioural detection and endpoint analytics tools are still effective at identifying and stopping AI-authored malicious code before it can execute. This suggests that while the methods used to create malware are changing, the fundamentals of good cyber defence—monitoring for suspicious behaviour rather than just known malware signatures—remain relevant and effective.

For small and medium businesses, this is an important reminder that keeping endpoint protection and monitoring tools active and up to date matters more than ever, even as threats become more sophisticated. Attackers may be using new tools to write malware faster, but the way that malware behaves once it tries to run is still detectable by modern security solutions.