- Hudson Rock describes one criminal marketplace that attaches owner profiles to stolen infostealer logs. The seller says the profiles are AI-made. Buyers can search them in plain language.
- Hudson Rock has not verified the shop's inventory or any completed sales. It says the profiles still need validation but could cut the manual work of choosing targets.
- Defenders should ask what an exposed machine did, what it held, and which sessions need revoking.
Stolen logs are familiar. Explaining them to the buyer is the new part.
A stolen password list is a pile of keys with no addresses. A buyer still has to work out which key opens something valuable. One marketplace now tries to answer that question for the customer.
Hudson Rock, a security firm that tracks infostealer malware, reviewed the service. It is advertised on cybercrime forums. Infostealers are programs that quietly copy saved passwords, cookies and autofill data from an infected computer. Each copy is called a log. In this shop, each log comes with a short profile of the machine's apparent owner, workplace and software.
The seller says the profiles come from AI analysis. Hudson Rock says the screenshots do not show which model is used or how accurate it is.
How the search works
Buyers type what they want in everyday language. One screenshot shows the request "citrix access to healthcare provider." The shop answered with a $10 log named after a clinic's front desk. Its profile points to a Canadian practice. It mentions medical software, billing work and saved Citrix remote-access logins.
Buyers can also narrow results by place, internet provider, malware family, operating system, date and price. The effect is a store catalog for compromised computers.
Other listings follow the same pattern. A $9 log is described as a defense contractor's household computer. The shop gives it a score of 100 out of 100. The screenshots show no method for the score.
What Hudson Rock checked
Hudson Rock matched several advertised logs to infections already in its own database. For the clinic log, a Firefox profile code in the shop's file preview matched a cookie-file path in Hudson Rock's records. Those records held Citrix-related login addresses and billing-related usernames.
Hudson Rock then asked Enki, its own AI analysis agent, about the owner. Enki pointed to a "Reception" username and to billing and reception addresses. It judged the role to be front-desk administration. The researchers add that several people may have shared the machine.
What remains unproven
The operator claims tens of millions of logs and over a billion parsed accounts. The operator also claims prices from $0.25. Those are the seller's numbers.
Alon Gal, Hudson Rock's CTO, told Help Net Security that the firm has not checked the full inventory. It has also not confirmed any completed purchase. Its firm finding is narrower: the stolen material is real, and it is offered with prices and payment tools.
Individual claims need the same care. One listing mentions "valid Anthropic Claude API session tokens." That is the seller's wording, and the screenshot does not show whether the tokens work. Search results for uber.com and fbi.gov show only that the shop lists records tied to those domains. They do not show a breach of either organization.
If it works as advertised
If this service works as advertised, it suggests a shift in what buyers pay for. The raw data may matter less than someone explaining it. That is an argument, not a measured trend. This is one marketplace, and Hudson Rock says the screenshots do not measure any improvement.
Gal's view is that novices get help spotting valuable logs. Experienced buyers spend less time reading them. He also said the profiles still need validation.
Searchable criminal shops are not new. The U.S. Department of Justice said Genesis Market let buyers search by location and account type. AI-assisted victim analysis has been documented before: Cisco Talos described the VOLTA malware using an LLM to assess stolen material. Hudson Rock shows no evidence linking VOLTA to this shop, and no earlier case has been shown inside a marketplace listing. What differs here is a description of each machine's job.
Why fresh logs and live sessions matter
Gal said some of the logs Hudson Rock matched came from machines infected less than a day earlier. He added that finding a log in Hudson Rock's records does not make it old or recycled. He also said an infection being found does not mean the affected service has acted on it. Many services, he said, do not watch for these infections or reset exposed logins quickly.
A password reset alone may not close the gap. Hudson Rock points to Microsoft guidance on token theft. Stolen sign-in tokens can stay useful after a user has passed multi-factor checks. Sessions must be cut off as well.
What to ask your team
Ask whether your security team can see infostealer exposure for staff and shared machines. Ask how long it takes a finding to become action.
Hudson Rock suggests three steps. First, map stolen items to accounts, applications and the machine's business role. Second, clean the infected computer, reset logins from a trusted device, and revoke sessions and tokens. Third, review sign-in and audit records for unfamiliar sessions or account changes.
Possibly shared front-desk computers and household computers that mix corporate access with personal accounts deserve a specific look. The listings highlight examples of exactly those.
Criminals are labelling stolen machines. Defenders gain by reading the same labels first.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: Hudson Rock.





