🔍 Read the full analysis: What AI Agents Could Change About Hacking At Scale on ThorstenMeyerAI.com
Get audio and creator gear 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 reported that an operator used Cairn, an open-source AI-agent platform, to attack more than 1,200 online shops from late September through mid-October. The attackers’ own records, reviewed by Reuters and NorForge, indicate break-ins at websites of more than 100 companies, but the amount of data accessed or taken is unknown.
An operator used Cairn, an open-source AI-agent platform, to probe more than 1,200 online shops from September 25 through mid-October, and the attacker’s records indicate break-ins at more than 100 company websites, according to Reuters and security firm NorForge. The reported victims include Deere, SharkNinja and Cummins; how much information, if any, was taken remains unclear.
NorForge researcher Snorre Fagerland found exposed Cairn control panels while searching for agent dashboards accessible on the internet. NorForge says it identified more than 20 panels linked to likely malicious activity, with new panels appearing over time. Fagerland could observe some activity through the exposed interfaces. NorForge published its report on October 19; Reuters published an investigation the same day and independently reviewed the operator’s Cairn activity to identify companies and contact them.
NorForge grouped the activity into three clusters. The largest, called AIA0001, used Chinese-language task instructions. NorForge counted more than 1,100 attack projects across 14 panels in that cluster, directed mostly at the United States, Japan and Australia. Reuters reported that the campaign targeted online shops in the United States, Canada, Australia and Japan. The figures describe different measures: projects recorded on a subset of panels and sites reportedly targeted across the wider campaign.
NorForge measured successful AIA0001 projects as taking a little under 23 hours from start to goal, on average as described in its report. It estimated the campaign’s AI costs at roughly $5,000, or about $50 per successful intrusion, while warning that the estimate carries considerable uncertainty. These are the firm’s estimates, not an independently established accounting of the operator’s expenses.
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
How Automation Changes Attack Scale
The reported campaign points to a practical change in how cyberattacks can be run: one operator can coordinate multiple software agents to carry out repeated, time-consuming work across many targets. That does not mean the agents acted without human direction or that every attempted intrusion succeeded. It does suggest that attackers may be able to automate parts of reconnaissance and intrusion work, allowing a small operation to keep probing while its operator provides only brief instructions.
For businesses, the affected sites may include more than primary corporate systems. Reuters’ reporting and NorForge’s analysis describe attacks on online shops and the potential value of maintaining access to e-commerce or payment infrastructure. Support sites, third-party vendor websites and overlooked web panels can also create exposure. A compromise at one such site does not by itself establish that a company’s core network or customer data was accessed.
Automation also creates a workload problem for defenders: repeated, machine-generated activity can make it harder to separate a serious intrusion from routine alerts. At the same time, companies need to consider how they identify and control software agents with access to their systems. The reported figures are specific to this campaign, however; they do not establish how common agent-led attacks are across cybercrime overall.
Products Worth Considering
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
LG UltraGear 32GX850A-B 32-Inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Exposed Cairn Panels Revealed Activity
The investigation followed a separate report from Gambit Security in September. The firm said it recovered a criminal staging server containing three AI-agent tools: Hermes, Strix and Cairn. NorForge then searched for exposed dashboards associated with Cairn. The platform is an agent-coordination system, not a hacking product by design, and its author has disputed that it was made for harmful activity.
NorForge describes Cairn as a system in which agents share information and planned actions through a common state board. In the campaign, according to the firm, shop websites were ranked and checked automatically before being passed to agents. Reuters reported that the operator’s instructions were fewer than 100 Chinese characters. The exposed control panels gave researchers an unusual view of the activity, but the available records do not answer every question about the resulting intrusions.
The September Gambit report described a separate agent-driven campaign involving at least 27 companies. Gambit said at least 600,000 card records were taken from two of them. That earlier case shows another reported use of agent tools, but it is distinct from the campaign Reuters and NorForge described; its findings should not be treated as evidence that payment-card records were stolen in the later attacks.
“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
Reuters said it could not establish how much data, if any, was taken. The operator’s identity is not known; the source material describes a Chinese-speaking actor, not a confirmed identity or nationality. The reports do not establish an arrest or charge. NorForge’s activity counts and cost estimates are based on the records and panels it reviewed, and the firm itself described substantial uncertainty around the cost estimate.
Company statements give only partial detail. Deere told Reuters it had no indication sensitive information was accessed. SharkNinja said a standalone product-support site was hit and that the threat was contained. Cummins said it was investigating a potential attack on a third-party vendor’s website. Those statements do not establish whether information was accessed at every affected company, whether access persisted, or whether all reported targets were successfully compromised.
Company Reviews and Further Findings
Companies named in the reporting may continue investigating affected websites, reviewing vendor access and determining whether data was accessed. Deere, SharkNinja and Cummins gave Reuters their current assessments, but the available reporting does not set out a common timeline for further updates or disclose the full scope of each company’s investigation.
Additional findings may depend on analysis of exposed Cairn panels, company incident reviews and any later disclosures from affected organizations or investigators. Until that work is public, the confirmed scope of the campaign is narrower than the possible consequences: researchers and reporters identified a large run of automated attacks and reported intrusions, but the amount of data taken and the operator’s identity remain unresolved.
Key Questions
What happened in the reported campaign?
Reuters and NorForge reported that an operator used Cairn AI agents to probe more than 1,200 online shops from late September through mid-October. The attacker’s records indicated successful break-ins at websites of more than 100 companies.
Who was behind the attacks?
The operator has not been identified. The reporting describes a Chinese-speaking actor, which does not establish the person’s identity or nationality.
Was customer or company data stolen?
That remains unknown. Reuters could not establish how much data, if any, was taken. Deere said it had no indication sensitive information was accessed; other companies described their investigations or the affected sites.
Is Cairn a hacking tool?
Cairn is an open-source platform for coordinating AI agents. Its author said it was not designed for hacking attacks; researchers reported that an operator used it in the campaign.
How much did the campaign cost?
NorForge estimated roughly $5,000 in AI costs, or about $50 per successful intrusion. The firm said the estimate was considerably uncertain, so it should not be treated as a verified accounting.
Source: ThorstenMeyerAI.com
