🔍 Read the full analysis: The AI Agent Swarm Behind A Hacker’s 100+ Website Targets on ThorstenMeyerAI.com
Get business pricing on office and shipping supplies
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Reuters, citing the attacker’s Cairn activity records, reported break-ins at websites of more than 100 companies during a campaign that began in late September. NorForge observed exposed agent-control panels and estimated the campaign’s AI costs, but the amount of data accessed or taken has not been established.
A Chinese-speaking operator used Cairn, an open-source platform for coordinating AI agents, in a campaign that tried more than 1,200 websites and broke into sites belonging to more than 100 companies, Reuters reported. The campaign began September 25, according to records reviewed by Reuters and security firm NorForge; how much information, if any, was taken remains unknown.
NorForge researcher Snorre Fagerland found internet-accessible Cairn control panels and monitored activity through them. The firm said it identified more than 20 panels it associated with likely malicious use, with new panels appearing during its observation. NorForge grouped activity into three clusters and published its findings on October 19; Reuters published its 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, mostly targeting the United States, Japan and Australia. Reuters separately reported that the operator’s records showed attempted targeting of sites in the United States, Canada, Australia and Japan, and identified website breaches at more than 100 companies. These figures describe different measures and sources: NorForge’s project count is not a count of confirmed victims.
Reuters named Deere, SharkNinja and Cummins among affected companies after reviewing the attacker’s Cairn activity and contacting organizations. Deere told Reuters it took “necessary remediation actions” and had no indication 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 attack on a website operated by a third-party vendor. Reuters could not establish whether data was taken or how much.
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 Expands the Attack Surface
The reported campaign illustrates how an operator can use agents to carry out extended, repetitive reconnaissance and testing across many websites. NorForge measured successful AIA0001 projects as taking a little under 23 hours from start to goal on average, with about 140 recorded “facts” per project. Those figures describe activity in the observed projects; they do not establish that every target was breached or that the system could operate without human direction.
NorForge estimated the campaign’s AI costs at roughly $5,000 in total, or about $50 per successful hack, while warning that the estimate carried “considerable” uncertainty. If the estimate is broadly accurate, relatively low operating costs could make prolonged, high-volume probing more accessible. The practical concern for companies is that overlooked entry points—including support pages, vendor-run sites and exposed administrative panels—may receive automated attention alongside primary business systems.
There is also a defensive workload issue. Large volumes of automated probes can produce more alerts and investigation demands, while security teams still need to determine which attempts led to access, what systems were reached and whether information was exposed. The reports do not establish that AI agents alone caused the breaches; they describe a human operator using an agent platform as part of the campaign.
Popular right now
KitchenAid Classic Series 4.5 Quart Tilt-Head Stand Mixer K45SS, Onyx Black
Ninja CREAMi Scoop & Swirl 13-in-1 Ice Cream and Soft Serve Maker, NC701
Sony BRAVIA 5 85-Inch Mini LED 4K HDR Smart TV (K-85XR50)
As an affiliate, we earn on qualifying purchases.
How Researchers Traced Cairn Activity
The findings emerged after a separate report by Gambit Security in September. Gambit said it recovered a criminal staging server containing three agent frameworks: Hermes, Strix and Cairn. NorForge then searched for exposed web dashboards associated with Cairn, using internet-search tools that index publicly reachable servers. The firm said most panels require authorization, but some it found did not.
NorForge described Cairn as a general-purpose, state-based problem-solving engine rather than a hacking product. Its workflow lets agents record discoveries and proposed next steps on a shared state board, so other agents can build on earlier work. The operator supplied tasks and could add information, while the system coordinated steps over time. NorForge’s report did not name victims; Reuters said it independently examined the attacker’s Cairn activity to identify companies and sought comment from them.
The Cairn campaign is not the only reported case of agents being used in cybercrime. Gambit Security’s September report described a separate campaign involving at least 27 compromised companies and at least 600,000 payment-card records taken from two of them, according to the firm. That earlier finding is related context, not evidence that payment-card data was stolen in the campaign Reuters and NorForge examined.
““Necessary remediation actions” were taken, and the company had “no indication that sensitive information was accessed.””
— Deere, in a statement to Reuters
Data Access and Operator Identity
The operator’s identity has not been publicly established in the source material. NorForge characterized the activity as Chinese-language, and the reporting described the actor as Chinese-speaking; that does not identify a person, establish nationality or show who directed the operation. The material also reports no arrest or criminal charge.
The extent of access at the named companies remains unresolved. Reuters could not determine how much data, if any, was taken, while Deere said it had no indication sensitive information was accessed. The reports do not provide a complete list of affected organizations or independently confirm every breach indicated by attacker records. NorForge also described its cost figure as highly uncertain, so the estimate should not be treated as a verified accounting of the campaign.
Company Reviews and Further Findings
The companies’ investigations and remediation work will determine whether any data or connected systems were affected. Cummins said it was investigating the potential attack on its vendor’s website; Deere and SharkNinja described response steps to Reuters. The reporting does not state a timetable for final findings or announce further law-enforcement action.
NorForge’s monitoring of exposed Cairn panels and the companies’ reviews may produce additional information about the campaign’s scope. For now, the clearest published account rests on the platform activity observed by NorForge and records Reuters reviewed. Whether those records correspond to further confirmed victims, and whether any access led to data theft or later misuse, remains open.
Key Questions
What happened in the reported campaign?
A Chinese-speaking operator used Cairn AI agents to target websites beginning September 25. Reuters reported attempted attacks on more than 1,200 sites and breaches at websites belonging to more than 100 companies, based on the operator’s activity records.
Which companies were named?
Reuters named Deere, SharkNinja and Cummins among affected companies. Their statements described remediation, a contained threat at a standalone support site, and an investigation involving a vendor-operated website, respectively.
Was customer or company data stolen?
The available reporting does not establish that. Reuters could not determine whether data was taken or how much. Deere said it had no indication sensitive information was accessed; the broader question remains unresolved.
Who was behind the operation?
The operator has not been publicly identified in the source material. The reports describe Chinese-language instructions or a Chinese-speaking actor, but do not establish the person’s identity or nationality.
Was Cairn designed as a hacking tool?
No such purpose is established in the reports. NorForge described Cairn as a general-purpose agent platform, and its author said it was not designed for hacking. The reporting concerns how an operator used the platform.
Source: ThorstenMeyerAI.com
