At about 0:55 a.m. on November 13, an intruder got into KARTE, the marketing platform where Japanese companies from eyewear retailers to MUFG Bank keep copies of their customer data. PLAID, the Tokyo company running it, detected the intrusion at 8:59 a.m. Japan Standard Time (JST). The intruder kept reading data until about 14:54, according to the times the company gave its customer MeetsMore. KARTE has been offline ever since.
The company has not said how many of its customers were affected. As of the morning of November 19, the Japanese security incident blog piyolog had found 78 affected organizations. Their reported effects range from leaked customer data to stopped services. Based on the highest record counts 23 of these companies published, up to about 96 million records may have leaked. That is a count of records, not of people.
The breach exposes two failures beyond the server flaw. The first failure is retention. The Tokyo company kept customer data after customers left, in at least one case from a contract ending in 2019. The second failure is still open. The intruder stole the secret keys and access tokens connecting KARTE to its customers' other systems. Any stolen key a customer has not yet revoked can still open that customer's connected service, a system PLAID never ran.
PLAID's contract and Japan's privacy law put much of the legal exposure on its customers. The archived version of KARTE's terms caps the company's damages at one month of the basic fee, unless it acted willfully or with gross negligence. Japan's privacy law makes each customer answerable for supervising PLAID. The amendment Japan's Diet passed in July adds no surcharge for a security failure like this one.
PLAID, headquartered in Chuo, Tokyo, has named no threat actor. Its notices and the KARTE status page are the only first-hand record of the intrusion.
Operation Timeline
Trace the events, turning points, and connections behind the analysis.
KARTE Intrusion Timeline
Key events from the KARTE terms of April 1, 2022, to PLAID’s status update on November 19, 2026.
View Timeline Hide Timeline
-
April 1, 2022
-
February 1, 2023
-
November 2023
-
January 31, 2025
-
June 2025
-
June 2026
-
July 17, 2026
-
November 13, 2026
-
November 13, 2026
-
November 13, 2026
-
November 13, 2026
-
November 13, 2026
-
November 13, 2026
-
November 13, 2026
-
November 14, 2026
-
November 15, 2026
-
November 16, 2026
-
November 17, 2026
-
November 17, 2026
-
November 18, 2026
-
November 18, 2026
-
November 19, 2026
Inside PLAID's Servers
PLAID's November 16 disclosure lays out the break-in. In the early hours of November 13, someone exploited a vulnerability in a PLAID server. The intruder stole the credentials the company's systems use to connect to its databases, then read data out of those databases. The intruder then used the same stolen credentials to take over one of the company's own accounts and log in to the KARTE admin console.
One set of credentials opened both the customer databases and a PLAID account with admin console access. Those are two of the most sensitive assets a platform holds, and nothing in the company's account suggests a second barrier between them. In MITRE ATT&CK terms, the intrusion ran through Exploit Public-Facing Application, Unsecured Credentials, and Valid Accounts.
PLAID has said nothing about how its server got popped. It has not named the vulnerable product, the flaw, or the exploit. It has published no Common Vulnerabilities and Exposures (CVE) identifier and no indicators of compromise. A flaw handing an intruder database credentials often does it by exposing configuration files or environment variables, through bugs such as arbitrary file reads or remote code execution. The company has not said which, if either, applies.
PLAID's response left the intruder inside for hours after detection. The company says it invalidated the stolen credentials right after detecting the intrusion at 8:59. Yet according to the window it gave MeetsMore, unauthorized access continued until about 14:54. That is 5 hours 55 minutes after detection, inside an intrusion lasting about 14 hours.
Nothing in the company's account explains the gap. It does not say when it locked the hijacked account. Invalidating the stolen credentials would not necessarily log out an intruder already signed in to that account. PLAID began pulling servers only at about 14:27. It finished containment, including blocking traffic from the attack source, by about 15:23. The KARTE status page announced the shutdown at 14:44.
On November 15, the status page said the shutdown was meant to stop data from continuing to flow out. That means data could still have been leaving PLAID until the shutdown, hours after detection. The Tokyo company says no new unauthorized access has appeared since it cut off the attack source. It is reporting to the Personal Information Protection Commission, consulting police, and planning an external forensic investigation.
What Left PLAID
KARTE is a platform for customer analytics, messaging, and marketing on websites and apps, Impress Watch explains. Companies load their own customer data into it. KARTE uses that data to show each visitor tailored content and send targeted email. Banks, newspapers, an electric utility, a football league, and some of Japan's largest retailers all kept customer data there.
PLAID confirms three categories of stolen data. The first is 79,249 email addresses belonging to staff at its client companies who log in to KARTE. The second is the secret keys and access tokens connecting KARTE to other services. The third is part of the customer data companies imported into KARTE. Each company chose which fields to import, so what leaked depends on the company.
Retailers and the utility handed KARTE full customer profiles. DHC, the cosmetics company, reports the largest count, about 19.11 million records with names, email addresses, and birthdates. JINS, the eyewear retailer, puts up to about 18.29 million accounts at risk, with surnames, emails, and birthdates. Ain Holdings says up to about 5.98 million Francfranc member records with names and municipalities may have leaked.
Shikoku Electric Power says names and addresses of about 600,000 Yonden Concierge members leaked. Goldwin says names, addresses, birthdates, and the last store each customer used may have leaked. Uniform Next confirms names, phones, and company names leaked for up to 1,701,006 customers. SMS lists names, addresses, and phones for 191,039 records from one of its services.
Others sent mostly email addresses. CAMPFIRE, the crowdfunding platform, says email addresses and hashed user identifiers leaked for 5,819,779 members. RIZAP Group says up to about 4.32 million email addresses may have leaked, about 60,000 with names attached. Optage counts up to 1.42 million mineo records, some from customers who canceled as far back as January 2020. Morisawa says records of 23,957 Morisawa Fonts buyers leaked.
MeetsMore falls in both camps. For 60,000 people, a name, email, phone, or birthdate leaked. For 3,398,469 more, only an identifier and prefecture leaked. Impress Watch reports Shimamura confirmed about 7 million member records leaked.
The financial firms sent data unable to identify anyone by itself. MUFG Bank leaked about 6.5 million items of gender, age, and region, and says nothing in them lets a third party identify or contact a customer. Mitsubishi UFJ Trust and Banking sent identifiers generated from internet banking member numbers. Mitsubishi UFJ eSmart Securities sent account numbers and ages with no link between them.
WealthNavi sent an internal code, birth year and month, and prefecture for about 910,000 customers. Golf Digest Online counts up to 6,974,506 member records and Club Tourism about 4.06 million, with no names in either. Nikkei had processed its subscribers' emails and phone numbers so they cannot be restored.
The banks, investment firms, and newspaper sent the least, so the least sensitive data leaked from them. The retailers and the utility sent full profiles. The 96 million figure blurs that difference and overstates the number of people. It uses each company's maximum, can count one person more than once, and mixes possible leaks with confirmed ones. Impress Watch notes the total scale remains unclear.
Keys to Other Systems
The stolen keys are the part of this breach still in play. The company asked customers to invalidate them, warning that otherwise someone could get into the connected services and write or leak data there. That means an intruder holding a live key could push content through a customer's email or messaging service, or alter its customer records. It has not said how many customers had such connections, and the count remains under investigation.
MeetsMore heard from PLAID at about 23:34 on November 13 that its access key had been stolen. It invalidated the key at once and found no logs of unauthorized log-ins with it. WealthNavi cut its data link and invalidated its credentials the next day. Odakyu Electric Railway disabled the KARTE link for its Odakyu Points app.
Others switched off services while they checked the damage. RIZAP Group stopped some chocoZAP app banners and store opening and closure emails from November 13. Docomo Finance stopped members from logging in through its website from 15:45 on November 17.
No public source shows the intruder using any of the keys. But the company has not published which keys were stolen or confirmed all of them are revoked. An email key left live at a single affected company is enough to send phishing from a trusted sender, to customers already expecting mail about this breach. Vendor credentials have been turned on customers before. Stolen Fireworks AI credentials exposed license data on 149,346 people at Uber.
Data Kept After Customers Left
KARTE's terms required PLAID to delete a customer's personal data within 90 days of the contract ending. At least five former customers have now found the company did not. Their data sat on the same systems the intruder read, long after the business reason for holding it had ended.
- Kaitech, a spot-work service in nursing care and medicine, used KARTE from February 1, 2023 to January 31, 2025. On November 15, 2026, PLAID emailed Kaitech to say Kaitech's data had been stolen. That data had sat on the vendor's systems for 651 days after the contract ended, about 18 months past the deletion deadline. Kaitech's follow-up review found 27,530 worker email addresses leaked.
- Welcome, which runs the HAY Online Store, left KARTE at the end of June 2025. It says PLAID kept its data in breach of terms requiring complete deletion within 90 days. Up to 9,890 customer names and email addresses leaked. Welcome has asked the company for a detailed report on the causes and for complete deletion.
- overflow left KARTE at the end of November 2023. Up to 29,950 records from its Offers job platform were still on the vendor's server. overflow warns usage data may also have leaked, which could reveal users' job-change and side-work plans to third parties. It is demanding PLAID delete the data and prove it did.
- Stripe International moved STRIPE CLUB off KARTE in June 2026, but part of its linked data stayed with PLAID.
- Sankei Digital ran its ZBAT! site on KARTE from 2017 to 2019. It had to ask PLAID whether its old data was affected. The company checked and confirmed names and email addresses of ZBAT! users from that period leaked.
None of these companies could have seen from its own side that PLAID still held its data. Once a customer leaves a platform like KARTE, it loses its view into what the vendor still holds. A deletion clause in a contract is a promise, and nothing in the KARTE case shows any customer verifying it was kept. overflow's demand for proof of deletion comes after the theft, not before.
Access left in place after a business relationship ended has caused other breaches. Expedia was breached through Qubole cloud roles former customers never deleted. Threat actors are also working the authority MSPs granted and never took back, and the best-documented AI platform breaches trace to credentials earlier cleanups left valid. In each case, access outlived the business reason for it.
What PLAID Promised in Its Contract
Under KARTE's own terms, keeping a former customer's personal data past 90 days is a breach of contract. KARTE's website now returns an access-denied error, so I read the version dated April 1, 2022, archived in June 2022. Kaitech and Welcome both cite its deletion term, so the clause still applied when they left. Each quoted clause is my translation from the Japanese, naming PLAID and the customer where the terms say the Company and the Service User. Article 40 sets the deletion deadline.
When the usage contract ends, PLAID will handle the data in the service relating to the customer's websites and apps as set out below. (1) The personal information. Delete within 90 days after the usage contract ends.
Article 28 separately requires each party to erase personal data promptly and safely once its use ends with the contract. Article 33 keeps both duties in force after the contract ends. Under these terms, the company was still in breach of its deletion duty to Kaitech and Welcome, both of which left after April 2022, when the intruder struck on November 13. Article 32 sets the security bar for both parties.
The customer and PLAID must, within a reasonable scope, take organizational safeguards such as internal rules, human safeguards such as employee training, and technical safeguards such as access control, identification and authentication of those accessing, and prevention of unauthorized access from outside, to prevent leaks, loss, and damage of information learned through use of the service and otherwise keep it secure.
A single stolen credential set opening the customer databases and the company's admin account alike is hard to square with that clause. The second paragraph of Article 32 gives the audit right to PLAID alone.
When PLAID asks for a report on the state of the measures in the preceding paragraph, the customer will comply within a reasonable scope.
I found no clause in the terms giving customers the same right over PLAID. A customer wanting to check the vendor's controls, or confirm a deletion, had nothing in the standard terms to enforce it. Article 42 decides what the company owes.
Notwithstanding the preceding article and any other provision of these terms, PLAID's liability for damages relating to the service is limited to an amount equal to one month of the service's basic usage fee, and to damage arising directly and actually to the customer. This does not apply where PLAID acted willfully or with gross negligence.
Keeping Kaitech's data about 18 months past the deletion deadline is the clearest argument any former customer has for gross negligence. Article 28 shifts every leak claim onto the party breaching the personal data clause.
If the customer or PLAID breaches this article, the breaching party will resolve on its own every problem arising from use beyond purpose, disclosure, or leak of the personal information, including claims for damages from visitors to the customer's sites, and will hold the other party free of all responsibility.
The company can point to the cap in Article 42. A former customer facing claims from its own members can point to the deletion breach. The terms do not say how the two clauses fit together.
The outage gives current customers a second claim, and a small one. Article 42 also pays for downtime only when the company is at fault and KARTE is wholly unusable for 96 hours. KARTE has now been offline more than five days. Even then, the payout is capped at the basic fee pro-rated for the downtime, with lost business and lost profits excluded.
None of this binds a customer holding its own contract. Article 45 lets an individual contract override the standard terms where they conflict. Any customer with a negotiated contract has terms the public cannot see.
What Japan's Privacy Law Requires
Under Japan's Act on the Protection of Personal Information (APPI), the breach is not PLAID's problem alone. Quotations from the law and from the Personal Information Protection Commission (PPC) guidelines are my translations from the Japanese. Article 25 makes each customer answerable for supervising its vendor.
When a business handling personal information entrusts all or part of its handling of personal data to another party, it must exercise necessary and appropriate supervision over that contractor so the entrusted personal data is kept secure.
The PPC guidelines define supervision as choosing an appropriate contractor, concluding a contract, and tracking how the contractor handles the data. They also say a company should not give a contractor data the work does not need. The same guidelines give, as their first example of failing to supervise, a case matching the KARTE customers' position.
Entrusting personal data to an outside business without tracking the state of its security measures at contract signing and as appropriate afterward, after which the contractor leaks the personal data.
That describes the position of every KARTE customer unable to show it checked the vendor's controls. It describes the former customers most of all, since none appears to have confirmed the company deleted its data. Article 22 adds a separate duty.
A business handling personal information must, within the scope necessary to achieve its purpose of use, endeavor to keep personal data accurate and up to date, and to erase the personal data without delay once it no longer needs to use it.
Data from ZBAT! users, kept seven years after that contract ended, is hard to square with that duty. The same duty applies to the customers. Optage sent KARTE records of mineo customers who canceled as far back as January 2020.
Article 26 sets reporting deadlines. It requires a company to report a leak to the PPC. The PPC expects a preliminary report roughly three to five days after a company learns of the leak. A leak possibly committed for an unlawful purpose, such as this intrusion, must be reported in full within 60 days of discovery. For a leak learned of on November 13, that deadline falls in January 2027.
Both PLAID and each customer owe that report in principle, and they may file jointly. DHC and SMS plan to report to the PPC jointly with PLAID. Under the guidelines, a contractor is exempt once it notifies the client company responsible for reporting. Emails from the vendor to former customers such as Kaitech can shift the reporting duty onto companies that left KARTE years ago.
Reports are going to more than one regulator. overflow filed its report under the APPI with the Kanto Bureau of Telecommunications of the Ministry of Internal Affairs and Communications. Whether regulators treat the supervision failures as the customers' or PLAID's will decide who answers for the former customers' data.
A breach like this one carries no direct fine under the APPI. Article 148 lets the PPC recommend corrective measures for violations of Articles 23 to 26, which cover security, supervision, and leak reporting. The PPC can order those measures if a company ignores the recommendation and serious harm is imminent. Violating the order is a crime, carrying a fine of up to ¥100 million for a company.
This year's amendment leaves that gap open. Japan's Diet passed it on July 10, and it was promulgated on July 17. It introduces surcharges equal to the financial gain from serious violations. They cover only improper use, improper acquisition, sharing data with third parties without consent, and misuse of a statistics exception. A security failure at a contractor is not on the list. The amendment takes effect, in principle, by July 2028.
Who Was Behind It
PLAID says only that a server vulnerability was exploited. Its notices carry no indicators of compromise. Six days after the intrusion, no threat actor has claimed the theft in the sources I read, and none of the victim notices adds attribution.
The intrusion began at about 0:55, outside Tokyo working hours, and the company detected it about eight hours later. In that time the intruder moved from a server flaw to stored credentials, then to the customer databases, then to an account with admin console access. That sequence suggests an intruder ready to turn one server flaw into access across the platform, not a scanner grabbing whatever one machine held. I weighed three hypotheses.
Hypothesis 1, Data Theft for Extortion or Sale
The intruder took records belonging to every customer, not one company. Retail and utility records with names, addresses, phones, and birthdates sell well and feed fraud. The company warns of suspicious emails to KARTE account holders. JINS and MUFG Bank warn their customers about phishing exploiting the leak.
ShinyHunters moved from stealing databases into OAuth abuse and SaaS integrations against Salesforce customers, and its dedicated leak sites keep coming back. The KARTE theft took both kinds of target, customer databases and the keys to connected services. SourceNext disclosed a threat actor abused its member information API from October 9 through October 13, potentially exposing as many as 22.5 million records.
Against it, no dedicated leak site lists PLAID or KARTE in any source I read. A listing is an adversary claim attached to a countdown clock, built to pressure the victim. Qilin claimed the Asahi attack while Asahi was still confirming the theft. A private demand to PLAID would not show, and the company has not said whether it received one.
Hypothesis 2, Espionage Against Japanese Finance and Media
MUFG Bank, two other MUFG companies, WealthNavi, and Nikkei are among the victims. Their leaked data identifies no one by itself, a poor fit for targeting individuals. An intelligence service would get more from the admin console access and the stolen client staff emails. Those addresses belong to the people who log in to KARTE at banks, a newspaper, and an electric utility. GOSSAMER BEAR showed how such a list gets used, posing as one victim's employee to phish a second.
Against it, the intrusion ended loudly. A platform-wide shutdown and dozens of public notices burned whatever access the intruder held inside PLAID. A spy service after the staff list would also have little use for millions of retail records.
Hypothesis 3, Access for Later Use
Live keys could open customers' connected services, the way an access broker resells footholds. MeetsMore found no log-ins with its key. No public source shows any key in use. Against it, the intruder spent hours reading customer databases, work an access broker after footholds would not need.
The evidence leans toward Hypothesis 1. The theft covered every customer at once. The most exposed records are retail and utility ones, which suit fraud and phishing. No victim has reported use of the integration keys. The absence of any dedicated leak site claim is the strongest point against it.
Three developments would move the assessment. A dedicated leak site listing, or a ransom demand the company discloses, would settle motive. Use of a stolen key against a bank or utility system would point to Hypothesis 2 or Hypothesis 3. PLAID naming the exploited product would let other companies running it check whether the same intruder reached them.
I assess with low confidence the intrusion was financially motivated. The company has published no indicators, no ransom demand or claim appears in the sources I read, and the admin console access remains unexplained. I assess with high confidence phishing against the exposed customers is the nearest harm. PLAID, JINS, and MUFG Bank all warn of it, and the retail records carry names and email addresses.
PLAID Failed on Security and on Its Own Contract
An outsider exploited a flaw in one of the company's servers and took the credentials its systems use to reach the customer databases. The same credentials took over an account with access to the KARTE admin console. One theft gave the intruder customer data from across the platform and the console used to run it. KARTE's own terms had promised access control, authentication, and prevention of unauthorized outside access.
The company also broke its deletion promise. It kept personal data from at least five former customers. Four left after the 2022 terms set a deletion deadline, and one had a contract ending in 2019. Those terms cap what PLAID owes for both failures at one month of fees unless it was grossly negligent. They also let PLAID demand a report on a customer's safeguards and give customers no matching right.
Companies buy platforms like KARTE as marketing tools, and the KARTE customers' own notices treat it as a marketing service. The breach shows what such a platform actually holds. It holds full customer profiles, keys into email and customer systems, and data from customers who left years ago. One server flaw at one vendor reached 78 organizations.
Japan saw the same concentration in October, when a threat actor took down a cloud shared by 495 Japanese companies and governments. The people whose data leaked from KARTE never chose PLAID. Most could not have named it, a vendor behind them they never picked and cannot audit.
When such a vendor is breached, the customer sends the apology, as Japan Airlines did after the Amadeus breach. KARTE lands days after Nitori disclosed up to 37.62 million records, and after Japan committed ¥10 billion to an emergency cyber security package.
The fixes now sit with the customers, because KARTE's terms leave them nothing else. Send a marketing vendor only the fields it needs, as MUFG Bank and Nikkei did. Treat every integration key as a production credential and rotate it on a schedule. When a contract ends, get written proof of deletion, which overflow is only now demanding. Write the right to audit the vendor into the contract, since KARTE's standard terms gave that right only to PLAID.
PLAID still owes its customers three answers. It has not named the flaw or published indicators of compromise. It has not said which keys were stolen, so no customer can verify it has revoked every key the intruder took. It has not explained why data from a contract ending in 2019 was still on its servers. Until it answers, every current and former KARTE customer has to assume its keys still work and its old data never left the company's servers.
Analyst Annex
Evidence status as of November 19, 2026. The records cover the intrusion into PLAID’s KARTE platform on November 13, 2026, and the disclosures KARTE customers published about their own data through November 18. Each customer row reflects that company’s own notice, and counts are the figures each company published. The table holds the larger and more instructive cases among the 78 organizations piyolog counts publishing impact.
No indicators of compromise have been published, so the Indicators subsection is omitted. Expedia, Amadeus, Nitori, and the AI developer platform breaches appear under Related Activity as context only.
Confirmed marks evidence published by a victim, vendor, regulator, or government authority about its own systems, decisions, or investigation. Reported marks technical findings or attribution assessments published by a third party. Claimed marks statements by a seller or self-identified threat actor that remain unverified. Provisional marks an analytical mapping supported by reported behavior but not made by the original source.
MITRE ATT&CK Mapping
| Tactic | Technique | Procedure | Status |
|---|---|---|---|
| KARTE Intrusion, November 2026 | |||
| Initial Access | T1190 Exploit Public-Facing Application | Exploited a vulnerability in a PLAID server in the early hours of November 13. PLAID has not identified the vulnerability Source: PLAID | Confirmed |
| Credential Access | T1552 Unsecured Credentials | Stole the authentication credentials PLAID used to connect to its systems and databases. Where the credentials were stored is not public Source: PLAID | Provisional |
| Privilege Escalation | T1078 Valid Accounts | Used the stolen credentials to take over a PLAID account and reach the admin console Source: PLAID | Provisional |
| Collection | T1213.006 Data from Information Repositories: Databases | Read data out of the databases. PLAID confirms the theft of customer data imported through the linking-table feature, 79,249 KARTE account email addresses, and integration secret keys and access tokens Source: PLAID | Confirmed |
Entities
| Entity | Role | Status |
|---|---|---|
| Threat Actors | ||
| Unidentified KARTE IntruderThreat actor, motivation not established | Exploited a PLAID server vulnerability, stole connection credentials, read the KARTE databases, and took over a PLAID account. PLAID names no threat actor Source: PLAID | Confirmed |
| Organizations and Victims | ||
| PLAIDBreached vendor, Chuo, Tokyo | Operates KARTE. Detected the access at 8:59 on November 13, finished containment by about 15:23, and has no restart date Source: PLAID | Confirmed |
| DHCVictim, cosmetics company | Names, email addresses, birthdates, ages, gender, member identifiers, and newsletter settings for about 19.11 million records Source: DHC | Confirmed |
| JINSVictim, eyewear retailer | Confirmed member identifiers, email addresses, surnames, birthdates, ages, gender, postal codes, and prefectures leaked, for up to about 18.29 million registered accounts Source: JINS | Confirmed |
| ShimamuraVictim, retailer | About 7 million member records confirmed leaked, according to Impress Watch Source: Impress Watch | Reported |
| Ain HoldingsVictim, retail and pharmacy company | Up to about 5.98 million Francfranc member records with names, birthdates, and municipalities, and up to about 2.6 million Ainz & Tulpe and AYURA records with names, emails, and phone numbers Source: Ain Holdings | Confirmed |
| Golf Digest OnlineVictim, golf services company | Up to 6,974,506 member records, including duplicates Source: Golf Digest Online | Confirmed |
| Club TourismVictim, travel company | About 4.06 million internet members’ encrypted member numbers, customer numbers, gender, age, and birthdates Source: Club Tourism | Confirmed |
| J.LeagueVictim, football league | 2,383,073 ticketing records with internal identifiers, birthdates, gender, prefecture, and municipality Source: J.League | Confirmed |
| CAMPFIREVictim, crowdfunding platform | Email addresses and hashed user identifiers of 5,819,779 members registered by November 11 Source: CAMPFIRE | Confirmed |
| RIZAP GroupVictim, fitness and wellness company | Up to about 4.32 million email addresses, about 60,000 with names. Suspended some RIZAP and chocoZAP web services from November 13 Source: RIZAP Group | Confirmed |
| MUFG BankVictim, bank | About 6.5 million items of gender, age, and region, with nothing letting a third party identify or contact a customer Source: MUFG Bank | Confirmed |
| Mitsubishi UFJ Trust and BankingVictim, bank | About 230,000 items, including identifiers generated from internet banking member numbers Source: Mitsubishi UFJ Trust and Banking | Confirmed |
| Mitsubishi UFJ eSmart SecuritiesVictim, securities firm | About 60,000 account numbers and about 105,000 ages, unlinked to each other Source: Mitsubishi UFJ eSmart Securities | Confirmed |
| WealthNaviVictim, investment service | Internal identifier code, birth year and month, and prefecture for about 910,000 customers. Invalidated its integration credentials on November 14 Source: WealthNavi | Confirmed |
| MeetsMoreVictim, services marketplace | 60,000 people with a name, email, phone, or birthdate and 3,398,469 with an identifier and prefecture only. Invalidated its access key on November 13 Source: MeetsMore | Confirmed |
| Uniform NextVictim, uniform retailer | Company names, names, emails, phones, postal codes, occupations, and industries for up to 1,701,006 customers Source: Uniform Next | Confirmed |
| Shikoku Electric PowerVictim, electric utility | Names, email addresses used as log-in identifiers, and addresses of about 600,000 Yonden Concierge members Source: Shikoku Electric Power | Confirmed |
| SMSVictim, online services company | Three services affected, the largest with name, nickname, email, phone, address, and birthdate for 191,039 records Source: SMS | Confirmed |
| GoldwinVictim, apparel company | Names, addresses, birthdates, emails, event participation, and last store used may have leaked. Count under investigation Source: Goldwin | Confirmed |
| MorisawaVictim, font company | 23,957 records of Morisawa Fonts buyers Source: Morisawa | Confirmed |
| Odakyu Electric RailwayVictim, railway | Internal identifiers and gender from the Odakyu Points app. Disabled its KARTE link Source: Odakyu Electric Railway | Confirmed |
| OptageVictim, telecommunications provider | Up to 1.42 million mineo records, including some customers who canceled from January 2020 Source: Optage | Confirmed |
| Docomo FinanceVictim, finance company | Suspended member-site log-in through its website from 15:45 on November 17 Source: Docomo Finance | Confirmed |
| NikkeiVictim, newspaper | Digital subscribers’ emails, phone numbers, gender, and birth month may have leaked, processed so they cannot be restored Source: Nikkei | Confirmed |
| WelcomeVictim, former KARTE customer | Ended its contract at the end of June 2025 and says PLAID kept data in breach of its 90-day deletion terms. Up to 9,890 HAY Online Store records Source: Welcome | Confirmed |
| overflowVictim, former KARTE customer | Ended its contract at the end of November 2023. Up to 29,950 Offers records remained on PLAID’s server, and it is demanding deletion and proof of deletion Source: overflow | Confirmed |
| Stripe InternationalVictim, former KARTE customer for STRIPE CLUB | STRIPE CLUB stopped using KARTE in June 2026, but part of its linked data remained with PLAID Source: Stripe International | Confirmed |
| Sankei DigitalVictim, former KARTE customer | ZBAT! contract ran 2017 to 2019 and Net Shop 2022 to 2024. PLAID confirmed leaks of names and email addresses of users from those periods Source: Sankei Digital | Confirmed |
| KaitechVictim, former KARTE customer | Stopped using KARTE on January 31, 2025, with deletion due within 90 days. Its data remained and 27,530 worker email addresses leaked Source: Kaitech, Kaitech | Confirmed |
| Tools and Services | ||
| KARTECustomer analytics and marketing platform | Used for customer analytics, information delivery, and marketing on websites and apps, holding customer data companies link into it Source: Impress Watch | Reported |
| Vulnerabilities | ||
| PLAID Server VulnerabilityVulnerability, not identified | Exploited to steal system and database connection credentials. PLAID has not described it Source: PLAID | Confirmed |
| Government Bodies | ||
| Personal Information Protection CommissionJapanese data protection regulator | PLAID is reporting the breach to it and consulting police. DHC and SMS plan to report jointly with PLAID Source: PLAID, DHC, SMS | Confirmed |
Related Activity
The records below cover other vendor and third-party breaches, Japanese disclosures and Japan’s government response from the past two months, a SaaS-focused eCrime threat actor, and a campaign reusing one victim’s identities against another. None was observed in the KARTE intrusion.
| Item | Detail | Status |
|---|---|---|
| GOSSAMER BEAR Second-Victim PhishingCampaign, reported November 2026 | Posed as one victim’s employee to phish a second organization Source: jark.me | Reported |
| Fireworks AI Credentials at UberIntrusion, reported November 2026 | Stolen Fireworks AI credentials exposed license data on 149,346 people at Uber Source: jark.me | Reported |
| MSP Access AbuseCampaign, reported November 2026 | Adversaries are working the authority MSPs granted and never took back Source: jark.me | Reported |
| Thai Airways NotificationsDisclosure, reported November 2026 | Notifying passengers whose booking records sat exposed in Amadeus for three years Source: jark.me | Reported |
| Japan Emergency Cyber PackageGovernment response, November 2026 | Japan committed ¥10 billion to an emergency cyber security package after a succession of intrusions and data breaches Source: jark.me | Reported |
| Expedia BreachIntrusion, reported November 2026 | Breached through Qubole cloud roles former customers never deleted Source: jark.me | Reported |
| Amadeus BreachIntrusion, reported November 2026 | Reached at least five carriers before anyone noticed. Japan Airlines told 92,000 customers Source: jark.me | Reported |
| Nitori ExposureIntrusion, disclosed November 9, 2026 | Unauthorized access may have exposed as many as 37.62 million records Source: jark.me | Reported |
| ShinyHunters SaaS Integration AbuseCampaign, reported October 2026 | Evolved from stealing databases into OAuth abuse, SaaS integrations, and voice phishing against Salesforce customers Source: jark.me | Reported |
| SourceNext API AbuseIntrusion, October 9 to 13, 2026 | A threat actor abused the member information API, potentially exposing as many as 22.5 million records Source: jark.me | Reported |
| Shared Japanese Cloud TakedownIntrusion, reported October 2026 | A threat actor took down the cloud 495 Japanese companies and governments shared Source: jark.me | Reported |
| AI Developer Platform BreachesCampaign, reported November 2026 | Sixteen platforms disclosed breaches in five weeks, the best-documented tracing to credentials earlier cleanups left valid Source: jark.me | Reported |
Intelligence Gaps
| Gap | Why It Matters | What Would Close It |
|---|---|---|
| The PLAID server vulnerability | Other platforms cannot check for the same flaw while PLAID names neither the product nor the class of bug | PLAID’s forensic findings or a vendor advisory |
| Use of the integration keys | Secret keys and access tokens for connected services were stolen, and PLAID has not said how many customers had integrations or whether all were revoked | A PLAID count of affected integrations and log reviews by the connected services |
| Activity between detection and 14:54 | Unauthorized access continued for 5 hours 55 minutes after detection, and nothing public explains what the intruder did in that interval or through the admin console | PLAID’s forensic timeline |
| Number of organizations and people affected | piyolog counts 78 organizations publishing impact, PLAID has not disclosed the total, and company counts mix maxima, overlaps, and possible leaks | Final counts from PLAID or the Personal Information Protection Commission |
| Data kept after contracts ended | At least five former customers report PLAID still held their data, including Sankei Digital from a contract ending in 2019, and PLAID has not explained why or how many others are affected | A PLAID statement on data retained after contract termination |
| Who answers under Japan’s privacy law | Article 25 of the Act on the Protection of Personal Information requires each customer to supervise PLAID, and Article 22 asks businesses to delete data no longer needed. DHC and SMS plan to report jointly with PLAID. The commission can recommend and order corrective measures, and the 2026 amendment adds surcharges only for misuse, improper acquisition, and unconsented sharing | A Personal Information Protection Commission recommendation, order, or guidance on the KARTE case |
| KARTE’s current contract terms | The only readable version, dated April 1, 2022, requires deletion within 90 days, caps PLAID’s damages at one month of the basic fee unless it acted willfully or with gross negligence, and lets an individual contract override the standard terms | The terms in force on November 13 and any negotiated customer contracts |
| Threat actor and motive | No threat actor is named, and no ransom demand or claim appears in the sources read | Attribution in PLAID’s forensic report, a police finding, or a claim backed by data |