Three Bank Failures in One Week, and Only One Was a Breach: A DFIR Field Guide to Chase, n0n and Sageworks
Three Bank Failures in One Week, and Only One Was a Breach: A DFIR Field Guide to Chase, n0n and Sageworks If you only read the headlines last week, American banks were under attack. Chase went down. JPMorgan, Visa and
Three Bank Failures in One Week, and Only One Was a Breach: A DFIR Field Guide to Chase, n0n and Sageworks
If you only read the headlines last week, American banks were under attack. Chase went down. JPMorgan, Visa and Mastercard had data leaked. A lending platform serving thousands of banks shut itself off. A ransomware crew claimed five banks and a credit union.
It looked like one story. It was not. These were three different kinds of failure, plus a fourth story that is older than it looks. Telling them apart matters, because each one leaves different artefacts, demands a different response, and teaches a different lesson. Here is what happened, what is confirmed, and what we think about each one.
1. The Chase outage: a software bug, not a breach
On Monday, November 16, JPMorgan Chase customers could not log in, see balances, or complete transfers and wires. Reports on Downdetector climbed past 2,000 by late morning and peaked around 4,800 to 5,000 in the early afternoon. Chase acknowledged the problem on X at 12:48 p.m. and posted its explanation at 3:57 p.m. By about 6:30 p.m., transactions were posting normally again, and Chase said systems were operating normally the next day.
The cause, as far as every current report confirms, was a change-management failure, not an intrusion. Chase spokesman Joe Evangelisti described it as "a temporary issue related to a software bug in on-premise hardware systems." A source told CNN that an update to Chase's physical computer systems triggered the failure. The FBI told NewsNation the outage traced to a software release rolled out that Monday, with no signs of malicious or external activity. The Department of Homeland Security said it found no proof of a cyberattack.
Two corrections matter here.
First, the spillover was narrower than it looked. Bank of America, Citigroup, Wells Fargo, U.S. Bank, Truist, Capital One and Zelle all showed spikes in user reports that day, but Bank of America and Citigroup confirmed their own systems were functioning normally, U.S. Bank said its systems were "not directly impacted," and the Zelle network was fully operational throughout. The delays happened only where one side of a payment was a Chase account. Chase's own framing, accepted by the Associated Press: "This is not a systemic issue."
Second, the rumours were wrong. Claims that Iran caused it, that it was a coordinated cyberattack or a global blackout, and that the nationwide Emergency Alert System test on November 17 was connected, have no evidence behind them. The EAS test was scheduled by the FCC on September 16.
The DFIR take: this is a resilience case study, not a breach investigation. The artefacts that matter are the release and change record (what shipped Monday morning, to which systems, canary or full rollout), deployment pipeline and configuration-diff logs, transaction queue latency and error rates by channel, and the rollback timeline. The harder DFIR work is proving a negative: to sustain "not cyber-related," you need the absence of evidence across authentication logs, privileged-access logs, endpoint alerts, and exfiltration telemetry for the affected systems over the same window. The public record contains the conclusion, not that evidence.
And there is a blast-radius lesson. One bank's bug became a multi-bank event because of counterparty concentration. Chase handles an estimated 22 percent of Zelle transactions. The question for your own shop: map which of your payment flows depend on a single counterparty's processing, and what your fallback is when that counterparty, not you, is down.
For customers: money was delayed, not gone. Check for delayed postings rather than duplicate charges, ask the bank to credit any overdraft or late fees caused by the delay, and expect phishing that name-drops the outage. Verify any "Chase" contact through the number on the back of your card, never a link in an unexpected message.
2. The n0n vendor breach: the real breach, and it is separate from the outage
On the same day Chase went down, an extortion group calling itself n0n listed JPMorgan, Visa and Mastercard on its leak site, in a batch of 59 organisations, with a 48-hour countdown. When the countdown expired on November 19, it published the data.
JPMorgan confirmed the incident that day. Its statement said third-party service providers were the target of a cyberattack, and that "based on our investigation so far, we believe the attackers accessed certain marketing data held by those vendors." The bank was explicit on two points: there was no disruption to its services, and there was "no connection" to the outage earlier in the week. It said it would contact affected customers directly and offer complimentary credit monitoring.
What was actually published, according to Cybernews and Reuters reviews of the leaked files: about 16,000 customer email addresses, some with phone numbers and JPMorgan account IDs, plus more than 100 bank employee email addresses, in a 502 KB file. The records appear tied to the invitation-only Reserve card programme. Visa's exposure was more than 6,000 premium cardholders in Asia-Pacific dining programmes, through a third-party provider in that region. Mastercard's was about 2,000 entries of limited non-payment email addresses from a specific benefit programme.
What was not in the files, per those reviews: no card numbers, no passwords, no account balances, no transaction histories.
Now the caveats, and they are load-bearing. The attack method is not confirmed publicly. The vendors are unnamed by all three companies. No attack date, vulnerability, or entry technique has been disclosed. The published file is far smaller than n0n's own claim of hundreds of thousands of campaign interactions, and Cybernews notes it is unclear whether the 502 KB is a sample or everything taken. Reuters could not independently verify the authenticity of the data. JPMorgan has not confirmed n0n was the actor, nor the 16,000 figure. The one researcher assessment worth noting: Cybernews found the same file layouts across the Visa, Mastercard and JPMorgan datasets, suggesting they came from the same third party.
The DFIR take: small data, high targeting value. Email plus phone plus account ID, plus the knowledge that someone holds an invitation-only Reserve card, is a spear-phishing kit for high-net-worth targets. The employee emails may be the more dangerous half. They enable help-desk impersonation, credential stuffing and business email compromise against staff, and one compromised employee can open a path into internal systems.
The artefacts investigators should demand: the vendor marketing platform's access and export logs (who pulled the Reserve member directory and campaign data, when, from where), the API keys and tokens the vendor held, and a file-hash and row-count reconciliation between what the vendor held and what n0n published, to test whether the leak is a sample or the full set. On JPMorgan's side, the claim to prove is the boundary: logs showing no access to core banking or card systems.
For customers, especially Reserve holders: assume targeted phishing that uses your name, your card programme, even your postal code. Do not act on "Chase," "Visa" or "Mastercard" emails or calls about this breach. Go to the number on the card. Take the free credit monitoring if offered, but remember monitoring detects new accounts, it does not stop phishing. A credit freeze is the stronger step, and it is free. There is no evidence here that justifies replacing cards, since no card numbers were in the files.
3. Abrigo Sageworks: a vendor pulled its own plug
On Friday, October 30, Abrigo detected "irregular activity" in its environment and disabled its Sageworks lending platform out of caution. Most of the platform returned on Monday, November 2. But borrower-facing features were still unavailable on November 16, while an outside firm completed platform testing. As of November 19, there is still no confirmation of whether any data was taken, and no final restoration date for the borrower portal.
Sageworks is the lending analytics and loan-application platform Abrigo sells to banks and credit unions. Abrigo serves more than 2,400 institutions. Sageworks holds the documents behind commercial and small-business loans, including tax returns, financial statements, rent rolls and credit reports. So the potential exposure is sensitive, even though no taking has been confirmed.
The attack method is not confirmed publicly. No vulnerability, phishing, credential or insider theory has been published. No group has claimed the attack. Abrigo told American Banker the unauthorised access was contained and there was no evidence of continued access.
One thing not to do: this is Abrigo's third security incident in about 14 months, and the first two involved business contact information tied to a Salesforce database. There is no public evidence connecting those Salesforce incidents to the Sageworks incident, so do not merge them.
The DFIR take: a vendor pulling its own plug is itself a signal. As Corsica Technologies CISO Ross Filipek put it, "Disabling a compromised account is one thing. Stopping a service customers depend on carries a different cost." Abrigo judged continued operation too risky while investigating. The question is what the "irregular activity" was: anomalous logins, unusual API calls, unexpected data staging, endpoint alerts? None of that is public.
The artefacts that matter: identity-provider and VPN logs for Sageworks admin and privileged accounts, Sageworks access logs showing who touched which borrower files, API-key usage, and data-integrity checks before the borrower portal returns. Researchers cited by BleepingComputer advised customers to rotate Sageworks API keys as a precaution. That is the right call even unconfirmed, since API credentials can register borrowers, obtain login tokens and launch portal sessions.
For banks using Sageworks: test the full borrower journey before resuming routine processing, reconcile every application handled manually during the outage against the platform, and ask Abrigo in writing which services are fully restored, whether data-integrity checks are complete, whether any borrower data was accessed, and how you will be updated. Keep the manual-workaround paper trail. It is both the recovery record and, if data was taken, the scope evidence.
For borrowers: if your bank used Sageworks, you may not know. Ask your lender whether your application was in the platform between October 30 and November 2, and whether they have received any notice from Abrigo. Watch for phishing that references a loan application you actually submitted.
4. Storm: the only confirmed SSN theft, and it is older and smaller than the headlines
Disclosed between November 6 and 11, a ransomware crew tracked as Storm, or Storm-1175, claimed five small banks and a credit union. Two have confirmed breaches. The other four have not confirmed anything, and none appears in the searchable breach records of the states checked. Do not list them as breached.
Sawyer Savings Bank in New York: a security incident in its managed service provider's environment on August 3. Branches closed for a week. Letters went to 16,069 people. Data taken: names and Social Security numbers. Sawyer said there was no evidence debit card numbers, financial account numbers, or core banking system information were involved.
Insight Credit Union in Florida: incident September 4 to 7, discovered October 13, reported to state attorneys general on November 6. The Texas filing lists 67,487 people. Data: Social Security numbers, driver's licence and other government ID numbers, financial account information, and credit and debit card information. Storm claimed on September 15 that it had taken 209 GB. That claim has not been confirmed by the credit union and has not been independently verified. Note the timing problem: the claim predates Insight's stated discovery date by about four weeks. That inconsistency matters for dwell time and for whether the claim and the filed incident are the same event.
The suspected entry point: Microsoft Threat Intelligence says Storm-1175 is "likely exploiting" a flaw in N-able N-central, the remote monitoring and management software that managed service providers use to administer client computers. The flaw is CVE-2026-18577, an unauthenticated authentication bypass, added to CISA's Known Exploited Vulnerabilities catalogue on August 3. But Microsoft has not confirmed that entry point, and neither Sawyer nor Insight has named it. Treat it as a hunt hypothesis, not a finding.
The DFIR take: the MSP and remote-management pattern is the story. One compromised N-central console can administer every client endpoint it manages. For a small bank that outsources IT, the managed service provider is the perimeter. The artefacts: N-central server authentication logs and patch version, remote-management session and command history (which commands ran on bank endpoints, by whom, when), and lateral movement evidence.
For customers of Sawyer or Insight: this is the highest-harm data set in the window, Social Security number plus government ID plus account and card numbers together. A credit freeze with Equifax, Experian and TransUnion is the strongest step, and it is free. Consider an IRS Identity Protection PIN and filing taxes early. For Insight members, the absence of a public credit-monitoring offer is itself something to ask the credit union about directly.
The thread that ties them together
Three different failures in one week, and the common thread is not an attacker. It is concentration: one counterparty on a huge share of payments, one vendor holding thousands of banks' lending data, one unnamed marketing supplier holding the richest customers' contact details, one managed-service console administering a small bank's whole environment.
That is the DFIR lesson worth carrying out of this week. When you scope an investigation, the blast radius is rarely the bank that called you. It is the vendor, the counterparty, and the console one hop away.
Sources
- Associated Press via TechXplore, November 17, 2026: Chase outage, software bug, not cyber-related. https://techxplore.com/news/2026-11-technical-issue-jpmorgan-triggers-online.html
- TheStreet, November 19, 2026: Chase outage, Zelle concentration, customer advice. https://www.thestreet.com/technology/jpmorgan-chase-outage-zelle-payment-delays
- The Hill via NewsNation, November 17, 2026: FBI, software release, no malicious activity. https://thehill.com/business/6226179-chase-outage-software-bug/
- Reuters, November 19, 2026: JPMorgan, Visa and Mastercard vendor cyberattack. https://wixx.com/2026/11/19/jpmorgan-says-third-party-vendors-targeted-in-cyberattack/
- Cybernews, November 19, 2026: n0n leak, about 16,000 JPMorgan Reserve emails, 502 KB file. https://cybernews.com/security/jpmorgan-chase-data-leak-non-ransomware-gang/
- DysruptionHub, updated November 19, 2026: Abrigo Sageworks timeline and status. https://dysruptionhub.com/incidents/profiles/abrigo-2026-01/
- eSecurity Planet, November 3, 2026: Abrigo shutdown, recovery checks, notification rules. https://www.esecurityplanet.com/news/news-sageworks-abrigo-suspicious-activity-outage/
- American Banker via ThreatCluster, November 11, 2026: Storm claims, Sawyer and Insight confirmations. http://threatcluster.io/article/ransomware-crew-claims-5-banks-and-a-cu-2-confirm-breaches-76a79bc0
- Emery Reddy, November 9, 2026: Insight Credit Union breach details. https://www.emeryreddy.com/blog/data-breach/insight-credit-union-data-breach
Originally published by Dev.to AI. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.