AI Agents Are Now Running Cyber Attacks at Machine Speed, Wiz Research Warns

Wiz Research says its incident response and threat hunting work shows a clear trend: attacks in which AI agents handle much of the hands-on work and may even orchestrate the intrusion. These agents use harnesses that write custom tools, check the results, adapt and try again at machine speed, while a human sets the goals and limits.

According to Wiz, these attacks range from sharply targeted to broadly opportunistic. An intrusion that starts with a single exposed service can grow into cloud credential theft, Kubernetes token theft, lateral movement (moving from one system to another) and bulk theft of secrets. AI agents shorten the skilled manual effort that this kind of escalation once needed. The attacks can also be verbose, with agents trying several approaches in parallel.

There is a potential upside for defenders. Wiz notes that current agentic attacks tend to be noisy and leave a broad trail of observable activity. The challenge is telling this activity apart from legitimate automation and linking individual signals into a clear picture of an attack. Wiz points to public disclosures involving named frameworks such as ARTEX and Hermes, as well as custom-built workflows and cases where the tooling was never identified. The post also reflects on what this shift means for attribution, attack scale and the future of defence. This summary is based only on the opening of the article.