🔍 Read the full analysis: How A Swarm Of AI Agents Could Reshape Website Hacking on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A Chinese-speaking operator used Cairn, an open-source platform for coordinating AI agents, to target online shops from late September into October, according to NorForge and Reuters. Reuters identified more than 100 companies whose websites were breached using records left on exposed control panels; how much data was accessed or taken is unknown.
An operator using the open-source Cairn platform directed AI agents to probe more than 1,200 websites from September 25 into October, with Reuters reporting break-ins at websites belonging to more than 100 companies. NorForge and Reuters based their accounts in part on activity records visible through exposed Cairn control panels; the amount of information accessed or taken has not been established.
Security researcher Snorre Fagerland of NorForge searched for internet-accessible Cairn dashboards and found more than 20 panels that the firm associated with likely malicious activity. The panels allowed him to observe agents’ tasks and progress. NorForge grouped the activity into three clusters and published its report on October 19. Reuters published a separate investigation the same day.
The largest cluster, labeled AIA0001 by NorForge, used Chinese-language task instructions. NorForge counted more than 1,100 attack projects across 14 panels, aimed mostly at sites in the United States, Japan and Australia. Reuters said it independently reviewed Cairn activity to identify affected companies and contacted them. The source account says Reuters reported targets in the United States, Canada, Australia and Japan; the precise scope and overlap between the datasets are not fully detailed.
Companies gave different accounts of their response. Deere told Reuters it had taken “necessary remediation actions” and had “no indication that sensitive information was accessed.” SharkNinja said a standalone product-support site was affected and the threat was contained. Cummins said it was investigating “a potential cyber-attack on a website operated by one of our third-party vendors.” Reuters could not determine how much data, if any, was taken.
One operator, a swarm of agents
The campaign in numbers
What the researcher and Reuters report about the busiest group NorForge found. Counts come from the attackers’ own exposed panels; cost and success figures are estimates.
The price tag estimates
NorForge / Fagerland, “uncertainty considerable”
- ~$5 per task on average, including tasks that fail early
- $100+ for long projects with more than 1,000 intents and facts
- ~$5,000 in AI costs for the campaign, about $50 per successful hack
- ~10% success rate, NorForge’s “reasonable assumption”
How the swarm runs
Simplified, no attack details
- Shop websites are ranked by traffic and checked automatically in a browser
- Each target becomes a Cairn project: a starting point and a goal
- Agents post facts and next steps on a shared board and build on each other
- Goal: a hidden way back in to shop and payment systems, for later
- 22 SepGambit reportGambitCairn found among three AI harnesses on a criminal’s staging server (separate campaign)
- 25 SepCampaign startsReutersA Chinese-speaking operator turns agents on shops in the US, Canada, Australia and Japan
- ≈ 2 OctOpen panels foundReutersFagerland starts finding Cairn panels in malicious use, about two dozen
- mid-Oct1,200+ sites triedReutersBreaks into websites of 100+ companies, incl. Deere, SharkNinja, Cummins
- 19 OctReport and investigationNorForge · ReutersNorForge publishes “Inside Cairn”; Reuters publishes its investigation
Sources: NorForge, Inside Cairn (19 Oct 2026); Reuters (19 Oct 2026); Gambit Security (22 Sep 2026). Numbers as published, same rounding as the article.
Powered by Thorsten Meyer AI
Automation Changes the Cost of Persistence
The reported campaign shows how an operator can delegate repetitive reconnaissance and testing across many targets rather than work through each site manually. NorForge measured a little under 23 hours from start to goal for successful AIA0001 projects. The firm estimated roughly $5,000 in AI costs for the campaign, or about $50 per successful intrusion, while warning that the estimate carries considerable uncertainty. Those figures are estimates, not verified accounting records.
If similar systems make sustained probing cheaper, security teams may face more activity to investigate and less time to distinguish routine automated traffic from an intrusion. The reported targets also include support sites and a third-party vendor-operated site, underscoring that a company’s exposure can extend beyond its main storefront or corporate domain. These implications are informed by the reported cases; the evidence does not establish that every company faces the same risk or that AI agents alone made the breaches possible.
The central security issue is not that Cairn is exclusively a hacking tool. It is a general agent-coordination platform that can be applied to different tasks. Its ability to keep shared task state and let multiple agents work in sequence can support legitimate automation, but can also assist an attacker who supplies harmful goals. Defenders may need to monitor not only people and devices, but also automated agents and the permissions they receive.
Products Worth Considering
Logitech MX Master 4 for Mac, Ergonomic Bluetooth Mouse with Haptic Feedback, Ultra-Fast Scrolling, USB-C Charging, for MacOS, iPadOS, Windo
Dell 34-Inch S3425DW Curved WQHD USB-C Monitor
Brother TD-2020A 2‑Inch Direct Thermal Desktop Label Printer with USB
As an affiliate, we earn on qualifying purchases.
Exposed Cairn Panels Revealed Activity
The reporting emerged after a separate investigation by Gambit Security in September. That firm said it recovered a criminal staging server containing three “AI harnesses”—programs that let AI models act with some autonomy—including Cairn. Gambit reported at least 27 companies compromised in that separate campaign and at least 600,000 card records taken from two companies. It is a distinct case, not evidence that those records came from the campaign Reuters and NorForge examined.
NorForge then searched for publicly exposed dashboards used by Cairn. According to its report, most such panels require authorization, but some did not have password protection. Fagerland found around two dozen panels he linked to malicious activity, with new ones continuing to appear. The attackers’ records supplied an unusually direct view of their operations, though the reports do not establish that every activity visible on those panels resulted in a successful breach.
Cairn’s author, who uses the name le1xia0, told Reuters that the platform “was not designed for hacking attacks.” The author added that harmful use was not their intention. That statement describes the developer’s position; it does not change the reported use of the platform in this campaign.
““Necessary remediation actions” had been taken, and there was “no indication that sensitive information was accessed.””
— Deere, in a statement to Reuters
Victim Impact Remains Unresolved
The operator’s identity has not been established in the reporting. NorForge described the activity as Chinese-language and attributed it to a Chinese-speaking actor; that does not establish the person’s nationality, location or identity. The source material reports no arrest or criminal charge.
It also remains unclear how much information, if any, was viewed, copied or removed from affected websites. Deere reported no indication of access to sensitive information, but Reuters could not determine the overall amount of data involved. The available records show what the agents attempted and what their operators logged; they do not independently verify every claimed outcome.
NorForge’s cost and timing figures are estimates based on the activity it observed, and the firm cautioned that uncertainty around the cost estimate is considerable. The reports do not provide a complete accounting of remediation costs, business disruption or losses across the companies involved.
Companies Continue Incident Reviews
Companies that identified affected sites will need to complete investigations, close access routes and assess whether customer, payment or business data was exposed. Cummins said it was investigating the vendor-operated website incident. The source material does not give a timetable for any company’s review or describe a planned public update from all affected organizations.
NorForge’s findings also point to exposed agent-control panels as a security concern: organizations using such systems may need to check that dashboards are not publicly accessible and that access is limited to authorized users. Further reporting could clarify the campaign’s full victim list, whether any information was taken, and whether investigators can identify the operator. Until then, the confirmed picture is of a large, automated probing effort and reported website intrusions, with the wider impact still under investigation.
Key Questions
What happened in the reported campaign?
According to NorForge and Reuters, an operator used Cairn-coordinated AI agents to probe online shops beginning September 25. Reuters reported break-ins at websites belonging to more than 100 companies.
Who was behind the attacks?
The operator has not been identified. NorForge described a Chinese-speaking actor based on the language of the tasks, but the reporting does not establish the operator’s nationality or location.
Was customer or payment data stolen?
The amount of data accessed or taken in this campaign is unknown. Reuters could not establish whether, or how much, information was removed. Deere said it had no indication that sensitive information was accessed.
Is Cairn a hacking tool?
Cairn is described as an open-source platform for coordinating AI agents, not a tool built specifically for hacking. Its author told Reuters it was not designed for attacks and said harmful use was not their intention.
How does this campaign relate to the September report?
Gambit Security reported a separate agent-driven campaign in September, including breaches at at least 27 companies and the theft of at least 600,000 card records from two of them. Those figures refer to that separate investigation, not the campaign described by NorForge and Reuters.
Source: ThorstenMeyerAI.com
