Hacker reportedly used AI agents against 1,200+ websites
Key takeaways
- A hacker reportedly used autonomous AI agents against more than 1,200 websites and compromised over 100 organisations.
- Researchers estimate the cost of a successful intrusion at approximately $50.
- The precise figures and extent of the compromises still need independent scrutiny.
More than 1,200 websites targeted, over 100 organisations compromised and an estimated cost of approximately $50 for a successful intrusion. Those are the figures attached to a reported campaign in which a Chinese-speaking hacker used autonomous AI agents, though the precise numbers and extent of the compromises still require independent scrutiny.
What was reported
According to the researchers investigating the campaign, much of the technical work was delegated to AI systems. Companies reportedly affected include Deere, SharkNinja and Cummins.
Treat the exact numbers as reported rather than settled. The researchers' cost estimate is the part that has drawn the most attention, because it suggests how cheaply an intrusion might be carried out.
Why it matters
Cybercrime has always benefited from automation. AI agents introduce something different: the possibility of automating substantial portions of an intrusion that previously demanded considerable technical expertise.
If these economics prove repeatable, organisations could face sophisticated attacks at a scale previously associated mainly with mass phishing and automated vulnerability scanning.
A wider pattern of supply-chain risk
The same digest points to other incidents showing how one failure can spread. Attackers reportedly exploited compromised credentials associated with ConnectWise Cloud Backup to reach Microsoft 365 customer environments, with reporting identifying 173 affected tenants. Malicious code was also reportedly introduced into Rank Math PRO version 3.0.123 packages distributed on 15 and 16 October. F5 released emergency security fixes for NGINX, including three critical-severity issues.
What to do this week
None of this needs exotic tooling. The ordinary defences are still the sensible starting point:
- Patch your CMS and every plugin, and remove the ones you no longer use.
- Turn on multi-factor authentication for every admin account. A hardware key such as the YubiKey 5 NFC is available on Amazon for the accounts you cannot afford to lose.
- Review third-party integrations and the privileged connections they hold into your systems.
- If you run the Rank Math PRO plugin, check the vendor's remediation instructions and verify the installed version.
We have been tracking this shift for a while. Our earlier pieces on an AI agent finding a Chrome exploit in six days, hidden commands aimed at AI in website headers and OpenAI's four rogue agent incidents show related trends from different angles.
The part worth watching is verification. Whether the approximately $50 figure holds up once other researchers examine the campaign will shape how seriously organisations treat this kind of automated attack.
Some links in this article are affiliate links. We may earn a small commission at no extra cost to you.
