Somebody built a hacking crew out of AI agents, pointed them at a piece of office printer software, and walked away.
Security researchers at GreyNoise found that a Russian speaking attacker used hundreds of AI agents built on OpenAI’s Codex and a DeepSeek model to break into print management software called PaperCut, the kind of boring backend software thousands of schools and offices run without thinking twice about it.
The numbers are the part that stop you. At least 440 systems compromised across 395 organizations in 48 countries. The attacker went from an empty workspace to breaking into a real victim in under four hours. Once the full campaign kicked off, the agents compromised eleven organizations in twenty six seconds.
Twenty six seconds. That is not a typo, that is just what happens when you hand the boring, repetitive parts of hacking to software that never gets tired and never needs sleep.
Most of the victims were schools. Education makes up roughly half of everything hit here, mostly because school IT departments are stretched thin and this kind of print software tends to sit forgotten in a corner of the network.
I am not going to pretend this is not unsettling. It is. Handing offensive hacking work to autonomous agents means one person with a laptop can now do what used to take a whole team.
But here is the flip side worth sitting with. The reason we know all this, the timeline, the exact IP addresses, the four hour mark, is because defenders are using the same kind of tooling to watch, trace, and expose this stuff fast. GreyNoise caught this pattern and published it within days, not months.
The same speed that makes attackers scary is available to the people protecting us too. That race is still very much on, and it is not one sided.
——
Follow: @Ali Demi
Book your free AI clarity call, NOW!
https://buff.ly/TpWy277
——
Sources:
https://www.greynoise.io/blog/ai-orchestrated-campaign-against-papercut-ng-mf
https://thehackernews.com/2026/09/papercut-attacker-uses-hundreds-of-ai.html
https://www.bleepingcomputer.com/news/security/ai-powered-attack-exploited-papercut-flaws-to-hack-395-organizations/
Repost this. Thanks.