🔍 Read the full analysis: The Price Of A Break-In: One Hacker, AI Agents, And 100+ Sites 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
Reuters, drawing on records from exposed Cairn control panels, reported that an unknown Chinese-speaking operator used AI agents to probe more than 1,200 websites and break into sites linked to over 100 companies. The reported campaign began Sept. 25; companies described remediation and investigations, while the extent of any data theft remains unclear.
An unknown Chinese-speaking operator used AI agents to probe more than 1,200 websites and break into sites associated with more than 100 companies, according to Reuters’ review of attacker records and research by security firm NorForge. The campaign, which began Sept. 25, targeted online shops and included websites linked to Deere, SharkNinja and Cummins; the amount of information accessed or taken has not been established.
NorForge researcher Snorre Fagerland tracked exposed control panels for Cairn, a free, open-source platform for coordinating AI agents. The panels were accessible on the public internet, allowing him to observe activity. NorForge identified more than 20 panels it linked to likely malicious use and sorted them into three activity clusters. Its report was published Oct. 19, the same day Reuters published its investigation.
The largest cluster, which NorForge labels AIA0001, used Chinese-language task instructions. NorForge counted more than 1,100 attack projects across 14 panels, aimed mainly at the United States, Japan and Australia. Reuters reported that the operator’s records showed attempts against more than 1,200 sites and successful break-ins at websites of more than 100 companies. NorForge’s report did not name victims; Reuters said it identified companies by independently reviewing the Cairn activity and contacting them.
Reuters reported that Deere said it took “necessary remediation actions” and had no indication sensitive information was accessed. SharkNinja said a separate 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 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
Why Automated Intrusions Matter
The campaign suggests that AI agents can carry out extended, repetitive probing with less direct operator attention than conventional manual attacks. NorForge measured successful AIA0001 projects as taking a little under 23 hours from start to goal, with about 140 recorded facts per project. One project continued for more than five days. Those figures describe activity in the observed records; they do not show that every attempt succeeded or establish the full damage.
NorForge estimated the campaign’s AI costs at about $5,000 overall, or roughly $50 per successful intrusion, while explicitly describing the estimate as highly uncertain. If the estimate is close, the low cost of sustained automated attempts could make it harder for defenders to distinguish routine activity from a targeted intrusion. The reported targets also include support and vendor-operated sites, not just a company’s main online store. A breach at a smaller or less closely monitored service can create exposure for a larger business, although the available reporting does not establish what access attackers retained or whether they monetized it.
Automated activity also creates a challenge for security teams: more alerts and repeated probes can consume time needed to investigate genuine incidents. The case raises questions about how organizations authenticate automated agents and monitor exposed management panels. Those are security implications drawn from the reported activity, not findings that every company in the campaign faced the same risk.
Products Worth Considering
Dell P2725H 27-Inch Full HD Monitor
Samsung Essential S3 27-Inch Curved FHD Monitor
Dell 24 Plus Monitor S2425HSM – 23.8 FHD (1920×1080) 144Hz 1ms, 2 x 3W Speakers, HDMI, Height/Tilt/Pivot/Swivel Adjustability, AMD FreeSync
As an affiliate, we earn on qualifying purchases.
How Cairn Panels Exposed Activity
The investigation followed a September report by Gambit Security on a separate agent-driven campaign. Gambit said it recovered a criminal staging server containing three systems that let AI models act with some autonomy, including Cairn. NorForge then searched for internet-visible Cairn dashboards, many of which require authorization. Fagerland found panels without passwords and began identifying ones linked to suspected malicious activity around Oct. 2.
Cairn is a general-purpose agent coordination platform, not solely a hacking product. In NorForge’s description, agents share a working record: each adds what it has found and what it intends to do, so another agent can continue from that state. The platform’s author, who uses the name le1xia0, told Reuters Cairn was not designed for hacking and that harmful use was not the author’s intention.
The earlier Gambit report concerned a distinct operation, not the same set of victims. Gambit reported at least 27 companies compromised and said at least 600,000 card records were taken from two of them. That separate case shows that agent-based criminal activity had been reported before the campaign Reuters examined; its figures should not be combined with the current campaign’s totals.
““It is a state-based problem solving engine, moving step by step in the general direction of a defined solution.””
— NorForge report
Data Access and Attribution Remain Unclear
The operator’s identity has not been established. The reporting describes the activity as Chinese-speaking, but that does not confirm who carried it out or where the person was located. The available sources do not report an arrest or criminal charge.
The victim count is based on Reuters’ review of attacker records, and the full extent of the campaign has not been independently confirmed for every company. Reuters said two smaller businesses acknowledged compromises; some contacted companies did not respond. The affected companies’ statements also do not settle whether data was accessed across all incidents. What information, if any, was taken, whether attackers maintained access, and whether any access was later sold or used remain unknown.
NorForge’s cost figure is an estimate, not a verified account of the operator’s spending. The source material says the firm considers the uncertainty considerable. The figures should not be treated as a reliable price for future attacks or as evidence that every intrusion had the same cost.
Company Investigations and Security Reviews
The immediate next step is for affected organizations to complete investigations, confirm what systems and information were exposed, and disclose material findings through their normal security and customer-notification processes. Cummins said it was investigating a possible attack involving a third-party vendor site. Deere and SharkNinja described steps taken in response, but the reporting does not provide a common timeline for further updates.
Security teams can also review whether internet-facing management panels are protected by authentication and whether third-party support or commerce sites are included in monitoring and incident plans. NorForge’s findings were based on control panels left exposed online, but the reports do not say how many organizations have since secured similar panels or whether the observed clusters remain active. Further reporting or company disclosures may clarify the scale of data access, the campaign’s status and the operator’s identity.
Key Questions
What happened in the reported campaign?
According to Reuters’ review of attacker records, an unknown Chinese-speaking operator used Cairn AI agents to probe more than 1,200 websites and break into sites associated with more than 100 companies.
Which companies were named?
Reuters identified websites linked to Deere, SharkNinja and Cummins among those affected. The reports do not provide a confirmed, complete list of every company whose site was compromised.
Was customer or company data stolen?
That has not been established. Deere said it had no indication sensitive information was accessed. Reuters could not determine how much data, if any, was taken across the campaign.
Who was behind the attacks?
The operator’s identity is unknown. Researchers described the activity as Chinese-speaking; that description does not establish the operator’s identity or location. The reporting cited here does not report an arrest or charge.
How much did the campaign cost to run?
NorForge estimated about $5,000 in AI costs for the campaign, or roughly $50 per successful intrusion, but said the estimate carries considerable uncertainty. It is not a verified cost figure.
Source: ThorstenMeyerAI.com
