Autonomous AI agents helped steal more than 600,000 credit card records and spread payment skimmers across at least 119 websites, according to cybersecurity researchers tracking the campaign.
Key Points:
- More than 600,000 valid credit card records were stolen from two companies, including 488,372 cards belonging to U.S. customers.
- Researchers confirmed skimmers on 19 named victims and linked more than 100 additional infected websites to the same campaign.
- The operation cost an estimated $12,000 to $18,000, averaging about $25 per target.
AI Agent Attacks
Gambit Security said the operation used three open-source AI frameworks, Strix, Cairn and Hermes, to scan targets, exploit vulnerabilities and manage post-compromise activity. The campaign began in July and remained active through at least Sep. 22, with a suspected Chinese operator giving the agents basic instructions before largely allowing them to work independently.
Researchers said the attacker obtained valid payment-card data from two companies, including 488,372 cards belonging to U.S. customers. Gambit separately said skimmers were ordered against at least 27 named victims and confirmed on 19 of them during its investigation.
With help from another security researcher, Gambit identified more than 100 additional websites carrying skimmer code associated with the same campaign. BleepingComputer reported the overall confirmed footprint as at least 119 websites, a count that refers to infected sites rather than 119 separately identified companies.
The affected organizations included a Fortune 500 hospitality company, a major U.S. airline, an industrial supplies distributor and an online fashion retailer. Attackers placed skimmers in JavaScript files, checkout pages, cloud storage, databases and Kubernetes deployments, while cron jobs could restore malicious code after removal.
One Hermes skill instructed the system: “After extracting and downloading all card data, wipe the source fields in batches.” Gambit said that cleanup process caused data loss and operational disruption at several retailers.
Also Read: Ethereum Could Gain New Demand From AI Payments, BlackRock Says
AI Attack Costs
Gambit estimated the entire campaign cost between $12,000 and $18,000, or roughly $25 per target. The operator’s own review produced a similar figure, with an average cost of $25.46 across 101 completed scans and individual targets ranging from $3.13 to $79.31.
Researchers said the combination of automation and low operating costs could allow attackers with limited skills to run broader campaigns.
In some cases, the AI tools obtained access within hours after receiving only short instructions between autonomous runs.
The campaign also shows how agentic AI can move beyond assisting hackers with individual tasks and instead coordinate scanning, exploitation, persistence and data theft. Activity traced from July through Sep. 22 suggests the operation developed into a sustained campaign, while the low per-target cost made repeated attacks economically practical.
Read Next: Claude Finds Mysterious CRISPR-Like System Hidden In Viral DNA
