← Blog
Data BreachThird-Party Riskn0nFinancial Services

Visa, Mastercard, JPMorgan breach: n0n hit the vendors

Visa, Mastercard and JPMorgan Chase confirm third-party vendor breaches after n0n leaked customer emails. What is verified, how n0n gets in, what to do.

Zero Hunt Research··7 min read

Published by Zero Hunt, an autonomous AI red team on an on-premise appliance running private AI: automated penetration testing for networks and infrastructure, black-box or gray-box, with a human approving every step that matters.

On November 19, three of the largest names in payments confirmed the same kind of incident on the same day. Visa, Mastercard and JPMorgan Chase all said a third-party vendor had been breached, after the n0n extortion group put customer data attributed to them on its leak site. None of the three says its own systems were touched, and so far no card numbers are involved. That is the reassuring half. The other half is that each bank's customer data sat at a vendor that a little-known group got into, most likely without a single exploit, and that customers now on a leaked marketing list are the natural next target for phishing.

As of 13:05 CET, 20 November 2026. The vendors have not been named; this post is updated as the companies disclose more.

What the three companies have confirmed

Company What it said (19 November 2026) Data involved Records in the leaked files (Hackread count)
Visa An incident "involving a third-party service provider in Asia Pacific"; "Visa's systems were not involved and were not impacted"; no payments data such as card numbers affected Not specified by Visa; the files hold customer emails, phone numbers and marketing campaign records, mostly from Asia-Pacific markets 7,815 unique email addresses
Mastercard "An incident at a third-party provider" exposing "a limited amount of email addresses" from "a small subset of cardholders enrolled in a specific benefit program"; no impact on the payment network; exposure contained Email addresses, "non-payments information" 3,210 unique email addresses
JPMorgan Chase Third parties providing services to the firm "were the target of a cyber attack"; attackers likely accessed "certain marketing data" that may include customer names and email addresses; customers to be contacted and offered free credit monitoring Marketing data, names, emails; Reuters' review found phone numbers and postal codes with some records, and about 124 employee addresses 16,541 unique email addresses

Sources: The Register for the Visa and Mastercard statements and Chase's post on X, SecurityWeek for the Reuters review of the files, and Hackread for the counts of unique addresses. Hackread itself notes the counts "do not establish how many customers were affected."

Two things the statements do not say. None of the companies has named the vendor or vendors, or said when the intrusion happened. And Chase's statement, as quoted, does not address card or account numbers either way, so read it as silence, not as a denial. Chase also said there is "no connection" between this incident and the service disruption some customers saw earlier in the week.

Who n0n is, and how it gets in

n0n appeared in September 2026 and moves fast. The Register counted 108 victims claimed in its first month, and SecurityWeek reports roughly 200 on the leak site now. It listed the three financial firms on November 16, set a short payment deadline, and published the files when nobody paid.

The more useful detail is how it works. In October, SOCRadar's threat hunters found an n0n operator's workspace on an open file server and documented 1,719 files, which The Register reported in detail:

  • No exploits. SOCRadar's CISO is quoted saying that "in every documented compromise in this investigation, access relied on weak or stolen credentials," with "no evidence of vulnerability exploitation."
  • The access methods were basic. Wide RDP scanning, automated credential stuffing against FortiGate SSL-VPN through a pool of residential proxies, an exposed AWS access key with read and delete rights on S3, and reused hosting-portal sessions.
  • An AI coding agent did the scripting. The operator ran OpenCode, an open-source coding agent, configured with 154 model entries across seven providers, some marketed as "uncensored". It was used to "write and debug hundreds of scripts", and the AI bill reached more than $7,216 by October 22. The researchers found no fully autonomous attack system: AI made a small crew productive, it did not replace them.

The attacker in this story is not a nation-state with a zero-day. It is a few people with a password list, a proxy pool and a $7,000 AI bill, working through vendors one login at a time.

We cannot say from public information that the vendors behind the Visa, Mastercard and JPMorgan data were breached the same way. The companies have not said. But it is the documented pattern of the group, and it is the one to test for.

Why the vendor is the soft spot

Marketing platforms, benefit-program operators and campaign agencies hold exactly what an extortion crew can sell or use: names, verified email addresses, phone numbers, program membership. They often sit outside the bank's own identity controls, behind a vendor's VPN or a SaaS admin console, protected by whatever MFA policy the vendor chose. Three facts in this case show the cost:

  • The data a bank considers "only marketing" is enough to send a convincing message to a premium cardholder who really is enrolled in that program.
  • Three firms with very large security budgets learned of the problem from an extortion listing, and none has yet said when the intrusion happened.
  • Customers do not distinguish "a vendor of the bank" from "the bank". The headline names Visa, Mastercard and JPMorgan.

What customers should do: expect phishing that knows your program

If you are a cardholder, the realistic risk this month is not card fraud from these files. It is a well-informed phishing message. The leaked records tie email addresses to specific programs and campaigns, so expect messages that mention a benefit you actually have.

  • Do not use links or phone numbers in a message about this breach, a "benefit", a refund or a locked account. Open your bank's app or call the number on the back of your card.
  • Treat a message that quotes your real program name as more suspicious, not less.
  • If Chase contacts you about credit monitoring, check that the offer arrives through a channel you can verify.

What security teams should take from it

For a bank, an insurer or any regulated firm, this is a third-party access review that should happen this week, not at the next annual vendor assessment.

  1. List which vendors hold customer contact data and how they authenticate to it: SSO with your MFA, their own accounts, shared admin logins, API keys.
  2. Ask each one three direct questions: is MFA enforced on every remote access path (VPN, RDP, SaaS admin), are any long-lived cloud keys in use, and when did they last rotate credentials for your environment?
  3. Check your own perimeter for the same pattern. n0n's methods work just as well against you: internet-exposed RDP, SSL-VPN portals that accept password-only logins, cloud keys in code or CI. Test them before a credential list does.
  4. Prepare the customer message now. If your data is at one of these vendors, your customers will get the phishing wave whether or not your name is on the leak site.

The regulatory frame already expects this. In the EU, DORA's testing and ICT third-party requirements make banks answerable for the security of the services they outsource, and NYDFS Part 500 requires a written third-party service provider security policy for New York regulated entities. "It was the vendor" explains an incident. It does not discharge the obligation.

What is not known yet

  • The vendors. None has been named, and it is not known whether one provider or several were involved. The regions in the files differ (Asia-Pacific for Visa; North America, the Middle East and Africa for the Mastercard files per Hackread), which suggests more than one.
  • When, and how many. No company has said when the intrusion happened or how many customers were affected. The address counts above come from the leaked files, not from the companies.
  • The entry point. Credential abuse is n0n's documented pattern, but nothing public ties these specific vendors to a method yet.

Where Zero Hunt fits

The question this incident leaves a security team with is evidential: can you show a supervisor, today, which outside parties can reach your customer data, how they authenticate, and that you tested it? Zero Hunt maps every finding and every retest continuously against 34 frameworks worldwide, including DORA, NIS2, NYDFS Part 500 and PCI DSS. Every record is signed when it is written, and reports are ECDSA-signed, so a third-party access review produces audit evidence rather than a spreadsheet of vendor answers.

The test itself matters as much as the paperwork. n0n got in with credentials, not exploits. Zero Hunt's autonomous AI red team, running on an on-premise appliance with private AI, includes a credential agent that tries the same paths against your perimeter. It checks exposed VPN and RDP services and default or weak passwords, with a human approving every intrusive step. Your data and credentials never leave the appliance.

Is this exploitable in your environment?

Zero Hunt answers that on your own network: an autonomous AI red team on an on-premise appliance, running on private AI, black-box or gray-box, with a human approving every step that matters. Proof of what is exploitable, the fix, and signed evidence — your data stays on your network.