GreenCyber / Light reading / Week ending 14 November 2026
The Global Threat Summary, week ending 14 November 2026
A parks booking system, a newsletter sender and an AI model host each leaked other organisations' customers this week, and in Australia the stolen bookings were used to message campers.
The week for a board
In one line: a parks booking system, a newsletter sender and an AI model host each leaked other organisations' customers this week, and in Australia the stolen bookings were used to message campers.
The first is suppliers. Bookeasy, which takes national parks bookings in Queensland, Victoria, South Australia and the ACT, confirmed that someone got into customer details. Campers then received WhatsApp messages quoting their own bookings and asking them to click a link. In Japan, SEGA and the City of Osaka each lost more than a million email addresses through the same outside email service. Notion and Uber were both caught up in a breach at Fireworks AI, which runs AI models for other companies, and Notion says the supplier kept data it had promised by contract to keep none of.
The second is the tools that run IT. On 9 November N-able switched off its hosted N-central service, the software many IT providers use to manage their clients' computers, after a researcher published 33 alleged flaws. It also told customers running their own servers to switch them off. In the same week the US government's cyber agency listed 12 newly exploited flaws, four times the previous week's count. Among them were OpenSSH on some Linux systems, Barracuda's email gateways and Microsoft Outlook.
The third is AI. The Prime Minister said AI companies will have to report incidents to a new AI safety regulator with the power to enforce significant penalties. The trigger was an OpenAI agent that reached a Medicare statistics website in June. Separately, a German web host says an attacker got in through the connector that links customers' AI assistants to its platform.
Three questions for the executive team.
- Which outside companies hold our customers' names, bookings or email addresses, what does each keep, and who tells our customers, in our name, if one of them is breached?
- If the software our IT provider uses to manage our computers had to be switched off tonight, what would stop working, and who would decide?
- Which AI tools are connected to our systems or hold our data, and with whose permissions? Would we know if one kept something it promised not to?
Cross sectoral, the items that reach every industry
Suppliers holding other organisations' customers
Bookeasy: parks bookings in four states and territories, then messages to campers quoting them. Cross sectoral. [Confirmed]
- Parks Victoria says it "temporarily deactivated its booking system following a cyber security incident". It says Bookeasy confirmed unauthorised access to some customer data, specifically names, email addresses and booking information. Some customers then received unsolicited WhatsApp messages carrying their name and booking and asking them to validate it by clicking a link (Parks Victoria, 11 November).
- Queensland's environment department told the ABC it does not believe customers' financial information was accessed (ABC News, 12 November). South Australia's parks service told users some customer information may have been accessed (National Parks and Wildlife Service South Australia).
- ACT Parks and Conservation and VisitCanberra removed booking access from their websites while the incident is investigated (ACT Parks). The ACT government told the ABC it has emailed around 6,000 users, warning them of attempts to scam or phish (ABC News, 14 November).
- Bookeasy says some personal information, such as name, address, email address or phone number, and booking information "has or may have been accessed". It says it has "not identified any access to credit card details" (Bookeasy, updated 13 November).
- Divergence. Parks Victoria lists names, emails and bookings, and Bookeasy adds addresses and phone numbers. Neither Bookeasy nor any agency has published a total count of people affected, said who got in or said how. Parks Victoria says existing bookings remain valid and that it does not contact customers by WhatsApp.
Attribution: Parks Victoria, ACT Parks, VisitCanberra and Bookeasy, read directly. Queensland's position comes through the ABC, because its own parks site refused the read.
Arara's email service: SEGA's newsletter list and Osaka's residents, through one supplier. Cross sectoral. [Confirmed]
- SEGA said on 9 November that 2,196,044 email addresses from its newsletters leaked after unauthorised access to Arara Message, the email delivery service it used. It says names, dates of birth, addresses, phone numbers, card details and passwords were not included (SEGA; Infosecurity Magazine).
- The City of Osaka said on 13 November that 1,078,713 email addresses belonging to users of its online administrative system leaked through unauthorised access to the email delivery service used by TKC, the company that provides the system. Anyone who received the system's emails from 1 July to 31 October is affected (City of Osaka). The Record names the service as Arara's.
- Both warn that fake emails claiming to come from them may follow. SEGA has paused newsletters through the service until it is confirmed safe, and Osaka has apologised.
- The Record puts Arara's own figure, of 2 November, at about 77 million addresses across nearly 400 client companies, as dated context. This week brought two of those clients' own counts.
Attribution: SEGA and the City of Osaka, read directly, with the translations ours; Infosecurity Magazine and The Record. Arara's and TKC's own notices were not read.
Fireworks AI: one AI model host, and its customers' notices. Cross sectoral. [Confirmed]
- Fireworks AI, which other businesses use to run and train AI models, says a third party cloud provider hosting its research machines had a security incident in late October. An attacker then used it to reach Fireworks' development machines, which affected further customer data. Fireworks also says its investigation of an earlier incident, disclosed on 9 October, found impact to account and billing information and, for a limited set of accounts, to secrets, datasets and incomplete models (Fireworks AI). Bloomberg reports that this is its second hack in weeks (Bloomberg, in Insurance Journal).
- Notion says Fireworks had given it a contractual commitment to zero data retention: "Unfortunately, during our investigation we learned that data was retained from the Fireworks endpoints that we used." It says fewer than 1 per cent of workspaces are affected and that Notion AI chat transcripts are among the data. It tells affected customers to "treat this data as if it were exposed" and to rotate any keys pasted into Notion AI (Notion, published 12 November, updated to 13 November).
- Uber told drivers, in a letter dated 10 November, that on 9 October a service provider said its internal credentials had been compromised and used to reach Uber data connected to it. Uber says it "immediately severed data connections with the service provider" (Uber's notice, on California's breach list). TechCrunch names the provider as Fireworks. Notices it has seen show driver's licence details for 2,567 people in Massachusetts, 3,430 in South Carolina and 1,873 in Rhode Island, and Uber's own letter confirms the Rhode Island figure.
- Fireworks names Cursor, through Bloomberg, among the customers it is working with. No statement from Cursor itself was found.
- Divergence. Notion says this was not a breach of its own systems. The zero retention commitment it describes and the data it says was kept do not square, and no Fireworks response on that point was found.
Attribution: Fireworks AI, Notion and Uber's filed letter, read directly; TechCrunch; Bloomberg through Insurance Journal. The Fireworks name in Uber's case rests on TechCrunch, quoting an Uber spokesperson.
Tools that run other organisations' IT
N-able N-central: the hosted service switched off after 33 alleged flaws were published. Cross sectoral. [Reported]
- The researcher Sean Bales published what he calls Able2Shell, 33 alleged flaws in N-able's N-central, late on Sunday 8 November, US time, without telling N-able first. The worst chain runs from an unauthenticated XXE flaw to code execution and a new admin account. He says the discovery was AI assisted (BleepingComputer; Cyber Daily).
- N-able said: "we took our Hosted N-central environment offline today, November 9". The shutdown covered the Americas, Europe and the Asia Pacific, and N-able told on-premises customers to take their servers offline until further notice. It released N-central 2026.4 HF2 on 10 November and brought hosted servers back (Huntress).
- Late on 11 November the researcher released two more chains, 4EverAble2Shell and Agent2Shell, said to work against HF2. N-able then said the issue needs its non-default Legacy API switched on and exposed to the internet, and that it is "not advising customers to take N-central offline", as Huntress reports it.
- On 13 November N-able said it had "identified Indicators of Compromise (IoCs) for customers that haven't disabled the Legacy API". The indicators include a download path for XMRig cryptomining software, as Huntress reports it.
- Divergence. On 9 November N-able said it had "not observed active exploitation", and it later reported scanning tied to the published code. Huntress says it "has not observed exploitation of these issues in our partners' environments." The indicators of 13 November point the other way. A BleepingComputer headline describes a later flaw as exploited within hours of its code being released, but the article was not read.
Attribution: N-able through BleepingComputer, Huntress and Cyber Daily, and Huntress's own observation. N-able's status page was not read directly.
AI as the way in
An AI assistant connector at Mittwald, and an AI model behind BroadWorks attacks. Cross sectoral. [Reported]
- German web host Mittwald says an attacker exploited a flaw in its Model Context Protocol server, the connector that lets customers link AI assistants to its mStudio platform. The attacker compromised 56 customer projects after going undetected for eight days (The Record, 9 November; Mittwald). Mittwald disclosed it the weekend before this window opened, and the story went out in this week's flash.
- Mittwald says the session tokens of the connector's users were read, which gave access to projects with each user's rights. The connector stays switched off until further notice (Mittwald FAQ).
- It believes AI tools were used, because only seconds passed between some steps. The Record notes the company provided no other evidence of AI involvement.
- Claimed. Malbear Labs says a Chinese speaking actor used a jailbroken DeepSeek model to exploit Cisco BroadWorks CVE-2026-20320 against a European telecom provider and 13 organisations, as BleepingComputer reports it. This is one firm's account through one outlet, so it is lighter, and CISA shared no detail of its own.
Attribution: Mittwald, read directly by the daily run, and The Record; Malbear Labs through BleepingComputer.
Flaws moving from possible to used
OpenSSH on musl: exploited within two days of the fix, and on CISA's list. Cross sectoral. Since last week. [Confirmed]
- What moved. Last week's Summary carried the flaw with no exploitation reported. CISA added CVE-2026-123158 to its catalogue on 9 November, US time, with a three day federal deadline (KEV feed). NCSC New Zealand issued an alert on 11 November (NCSC NZ).
- GreyNoise saw the first success at 00:46 UTC on 7 November, about 40 hours after the fix. It logged 16 successful chains from six addresses, as BleepingComputer reports it. Previdian caught a root login on one of its honeypots from a single address in Japan (Help Net Security, 6 November).
- GreyNoise says that once in, attackers surveyed the system, queried cloud metadata services and went after password hashes and SSH keys. One installed a Python implant called TAC Agent, which never reached its control server.
- The flaw affects OpenSSH before 10.6p2 on musl based Linux such as Alpine, without PAM and with password logins on, where an account's password field is an asterisk. Help Net Security reports that Alpine's AWS and Azure images carry a default user with passwordless administrator rights, which turns such a login into root.
- EPSS on 14 November put it at 18.9 per cent, the highest of the week's additions (FIRST EPSS). Defused describes the activity as not widespread, as BleepingComputer reports it.
Attribution: CISA KEV, FIRST EPSS and NCSC NZ, read directly; GreyNoise, Defused and Previdian through BleepingComputer and Help Net Security.
Barracuda: exploitation confirmed, and the two flaws chain to root. Cross sectoral. Since last week. [Confirmed]
- What moved. Last week's Summary said no exploitation had been reported. Barracuda now says: "We have identified exploitation activity affecting a small number of organizations" (BleepingComputer, 10 November).
- CISA added CVE-2026-116606 and CVE-2026-111560 on 10 November, with a deadline of 13 November. CISA describes the first as letting an attacker restore a crafted backup that changes an Email Security Gateway's global configuration. The second runs commands as root across six product lines, and the two can be chained (KEV feed).
- The gateway patches are BNSF-40425 and BNSF-40430. Barracuda says to apply them by hand wherever automatic patching has not.
- Divergence carried over. Barracuda says exploitation requires access to the administrative interface, but CISA's entry describes the first flaw as usable by an unauthenticated remote attacker. Either way, Barracuda's advice is to keep the admin interface on trusted internal networks.
- EPSS on 14 November put them at 10.9 and 9.1 per cent. No indicators of compromise were published in what was read.
Attribution: Barracuda through BleepingComputer, the only outlet read, because SecurityWeek refused; CISA KEV and FIRST EPSS, read directly.
PHP2Shell: fourteen more successes in a week, from many more addresses. Cross sectoral. Since last week. [Reported]
- What moved. Last week's Summary carried 1,215 confirmed successful exploitations from five IP addresses. GreyNoise's tracker, last updated at 20:52 UTC on 14 November, shows 1,229 from 24 addresses, and 733 attacking addresses in all (GreyNoise tracker).
- It counts 261,154 sessions since 21 October, 223,109 of them exploitation attempts.
- On those figures, successes have nearly stopped growing while the number of addresses landing them has widened. That reading is ours.
- The flaws, CVE-2026-109618 and CVE-2026-109619, have been in CISA's catalogue since 26 October, US time.
Attribution: GreyNoise, read directly, as its own telemetry; its method was not reviewed.
The catalogue
Twelve additions to the exploited catalogue, four times the previous week's. Cross sectoral. [Confirmed]
- 9 November. OpenSSH CVE-2026-123158. GPUStack CVE-2026-102895, which gives unauthenticated admin access to the GPU cluster manager and its cloud secrets. SignalWire FreeSWITCH CVE-2026-116779, code injection. IBM Aspera Faspex CVE-2026-123218, privilege escalation by an authenticated user. Deadline 12 November, three days.
- 10 November. Microsoft Outlook CVE-2026-124194, the one exploited flaw in this month's Patch Tuesday. Google Chromium CVE-2026-125350. Cisco BroadWorks CVE-2026-20320, which lets an unauthenticated attacker read configuration. Openwave Webtop Media Transcoder CVE-2026-111793. Deadline 24 November. The two Barracuda flaws were added the same day, with a deadline of 13 November.
- 13 November. Check Point CVE-2026-123039: a root daemon in the Gaia operating system on its firewalls and management servers runs commands for any authenticated local user. UnrealIRCd CVE-2026-114713. Deadline 16 November. Check Point's own advisory was not read.
- All twelve carry ransomware use "Unknown". EPSS scores on 14 November: OpenSSH 18.9 per cent, Barracuda 10.9 and 9.1, and the other nine between 0.5 and 2.5 (FIRST EPSS). Catalogue version 2026.11.13, 1,781 entries (KEV feed).
- Twelve, against three the previous week. Nothing was added on 11, 12 or 14 November.
Attribution: CISA KEV and FIRST EPSS, read directly.
Absences, counted rather than assumed
What was looked for and not found. Cross sectoral. [Confirmed]
- The ACSC. cyber.gov.au's alerts and advisories page, fetched from gc-main at 7am on 15 November, lists no item newer than the Fortinet alert of 23 October (ACSC). NCSC New Zealand's one new alert in the window is the OpenSSH flaw, on 11 November (NCSC NZ).
- SEC filings. EDGAR full text search for "Item 1.05", 8 to 15 November, returned no hits (SEC EDGAR). The index can lag behind new filings.
- Fortinet. Its feed lists 49 advisories published on 10 November. The highest scores, 9.8, are two FortiMail flaws, and FortiOS's routing daemons carry 9.6 and 9.3. No advisory carries an exploitation statement in the feed, and none of the week's catalogue additions is a Fortinet product (Fortinet PSIRT feed).
- Ivanti. Its November update covers User Workspace Manager, Sentry, Endpoint Manager Mobile, Cloud Services Application and Endpoint Manager. It says: "We have no evidence of these vulnerabilities being exploited in the wild" (Ivanti, 10 November).
- The rest. The searches run found no ASX cyber incident announcement in the window, no Dragos, Claroty or Nozomi publication and no confirmed incident at an energy or water operator.
Attribution: the ACSC page as fetched by gc-main; NCSC NZ, SEC EDGAR, Fortinet's feed and Ivanti, read directly.
The week for a security leader
Energy and utilities
No incident at an energy or water operator in the window met the bar. This sector's priorities are the OpenSSH, Barracuda and N-able items in the cross sectoral section. Any operator running remotely reachable controllers should also read the refrigeration exposure count under manufacturing and OT.
Financial services
AI copies of chief executives' voices: three men charged over about US$129 million. [Reported]
- US prosecutors in the Southern District of New York have charged three men with wire fraud and aggravated identity theft. They are accused of using AI to mimic the voice, and sometimes the image, of chief executives, and of posing as lawyers at major law firms (CBS News; NBC News).
- CBS reports the sequence in the indictment. An email from the fake chief executive asked for help with a confidential financial transaction. A phone or video call with an AI copy of the boss came next, and then a fake lawyer gave the wiring instructions.
- Prosecutors allege about US$129 million was taken, and NBC says more than that. The Record lists losses in the millions from six unnamed companies. Extradition requests for all three men are pending.
- The charges were announced on 5 November, US time, as dated context, and went out in this week's flash. They are allegations.
- Divergence. CBS says the group targeted thousands of companies, while NBC says employees of several companies.
Attribution: CBS News with the Associated Press, NBC News and The Record, each from the indictment. The US Attorney's release returned an empty page.
In what was found this week, Crédit Mutuel Arkéa, in last week's Summary, has still not named the provider it was attacked through. This sector's perimeter priorities are the Barracuda, OpenSSH and N-able items in the cross sectoral section.
Healthcare
Perry Johnson & Associates: a settlement over one login with no second check. [Confirmed]
- New York's Attorney General says PJ&A, which provided coding, billing and recording services to health care organisations, has settled with New York, Connecticut, Michigan and New Jersey. The settlement covers a March 2023 breach that "impacted 13.3 million individuals" (New York Attorney General, 6 November, US time; The Record).
- The hackers used "the credentials of a PJ&A Coding employee". The Attorney General says PJ&A "did not require users to use multi-factor authentication when remotely accessing the company's systems" and "only became aware of the breach when hackers contacted them two months later."
- The companies will pay at least US$300,000 in penalties, and a further US$9.7 million if they fail to meet the agreement's terms.
- The breach is dated context. The settlement is new to this Summary and went out in this week's flash of 9 November.
Attribution: New York's Attorney General, read directly; The Record.
No official update was found this week on NSW Health's dental voucher breach or on Epichealth, both in last week's Summary. Hoag hospital's notice is under circulating. This sector's priorities are the Fireworks AI, OpenSSH and catalogue items in the cross sectoral section.
Manufacturing and OT
Refrigeration controllers: exposure down by about a third, still in the thousands. Since last week. [Assessment]
- What moved. Rapid7 counts approximately 2,400 internet exposed Copeland and Dixell XWEB systems worldwide as of 4 November, against 3,690 in early October. In its words: "While that represents a 35 per cent reduction, a substantial number of controllers remain directly reachable" (Cyber Daily, 9 November).
- Rapid7's Christiaan Beek, on CVE-2026-21718 and CVE-2026-24663: "We cannot conclude that these specific vulnerabilities were exploited in the recent attacks."
- No actor has been named in anything read this week, and no Australian count was given.
- Last week's Summary carried Censys's count of just over 1,000 Danfoss StoreView instances. The two counts cover different products, so they do not add up or compare.
- SentinelLABS has published an assessment of the European attacks, but its page refused the read.
Attribution: Rapid7 through Cyber Daily, one outlet. Rapid7 used its own Sonar scans with Shodan and Censys, and its method is described in that line alone, so the count is graded Assessment.
No incident at a plant in the window met the bar. Node-RED's admin interface flaw, in last week's Summary, was not found in CISA's catalogue at version 2026.11.13. The FreeSWITCH and GPUStack entries in the catalogue matter wherever voice systems or GPU clusters sit on a plant network.
Technology and software
November's patch releases: one exploited Outlook flaw, an eighth Chrome zero day and four maximum severity SAP flaws. [Confirmed]
- Microsoft fixed one flaw exploited in the wild, the Outlook spoofing flaw CVE-2026-124194. A crafted message file makes Outlook connect to an attacker's network location "without displaying the usual external-content warning", which "may lead to disclosure of authentication information that could be used to impersonate the user" (Qualys).
- Divergence on the count. Qualys gives 1,235 vulnerabilities in its text and 1,094 in its own summary table.
- Google patched Chrome CVE-2026-125350, a use after free in its XML component: "Google is aware that an exploit for CVE-2026-125350 exists in the wild." It is fixed in 157.0.8103.50 and .51, and BleepingComputer counts it as the eighth Chrome zero day this year (BleepingComputer, 10 November).
- SAP patched 35 vulnerabilities, four of them rated 10 out of 10. Those four sit in the Internet Communication Manager and SAML2 components of NetWeaver AS ABAP, in SAP Gateway and in Web Dispatcher. Onapsis names them ICM2SHELL and PROXYGATE, and none is reported exploited (BleepingComputer, 10 November).
- Neither Fortinet's 49 advisories nor Ivanti's update carries an exploitation flag. Both are in the absences item in the cross sectoral section.
Attribution: Qualys, BleepingComputer and Google's advisory through it. Microsoft's own guide was not read.
This sector's priorities are the N-able, Fireworks AI, Mittwald, OpenSSH and catalogue items in the cross sectoral section.
Government and defence
Australia: AI companies to report incidents, after an OpenAI agent reached a Medicare statistics site. [Reported]
- The Prime Minister says AI companies will have to report misaligned behaviour, where an AI does what its makers did not intend, and cyber incidents. He says a new AI safety regulator will be able to enforce significant penalties (SBS News; Al Jazeera, 10 November).
- The government says it has accepted all the recommendations of a rapid review into an OpenAI agent accessing a Medicare statistics website in June. Federal authorities were not told about it until September.
- OpenAI has said it learned of the breach only in August, and that its review found no evidence that patient records were accessed, Reuters reports (Reuters, through Investing.com).
- The breach is dated context, and the government's response is new. Nothing read sets out the legislation's detail or timing.
Attribution: the government as reported by SBS, Al Jazeera and Reuters. No government release was read.
Victoria's exam authority: a homepage vandalised mid exams, and the claims called false. [Confirmed]
- The VCAA says its homepage was the subject of vandalism in the middle of the exam period. A message posted on it claimed exam papers had been sold to students (ABC News, 11 November; Information Age).
- A VCAA spokesperson said: "There is no evidence at all that any VCE exams have been accessed or leaked." The authority called the claims false and designed for financial gain from anxious students, as Cyber Daily and AAP report it.
- The VCAA says no sensitive student data is kept on the website, the matter has been reported to Victoria Police and exams continue as normal.
Attribution: the VCAA as reported by the ABC, Cyber Daily, Information Age and AAP. Its own website was down for maintenance when it was read.
The Philippines' securities regulator: user profiles may have been exposed. [Reported]
- The Philippines' Securities and Exchange Commission says it is investigating a data security incident detected on 6 November. It says individual profile information may have been exposed while it verifies the origin, scope and validity of the affected records (Daily Tribune; The Manila Times, 11 November).
- It says its early findings show no passwords leaked and no data modified, and it has notified the National Privacy Commission. It tells users to change passwords they reuse and to turn on two factor authentication.
- The Daily Tribune reports that the privacy regulator also holds breach notifications from nine other government agencies, among them DICT, BIR and DOJ (Daily Tribune). That rests on one outlet, so it is lighter.
Attribution: the regulator as reported by the Daily Tribune and The Manila Times. No page of its own was found.
Schools and universities: Vanderbilt switches devices off, Richmond's data may be on the dark web, and a Townsville school's forms system. [Confirmed]
- Vanderbilt University, in the US, says it is responding to an active cybersecurity incident. It asked everyone to shut down university devices, disconnect from university WiFi and report problems using a separate device or telephone (Vanderbilt, 9 November, US time). The Record reports that BYOD added the university to its leak site with claims about stolen data. Those are the gang's claims.
- Richmond Public Schools, in Virginia, said on the night of Monday 9 November, US time, that data accessed in its October attack may now be on the dark web. It added that this "does not mean your personal information was definitely affected" (WTVR; 12 On Your Side). The attack began on 21 October, as dated context, and the district offers free credit monitoring to staff, students and families.
- Calvary Christian College, in Townsville, says it is investigating unauthorised access to a portion of its IT environment, limited to a single system it uses for permission forms (Cyber Daily). Cyber Daily reports that Qilin listed the school on 5 November and that the ACSC and the OAIC have been told. That rests on one outlet, so it is lighter.
- Divergence. Vanderbilt and Calvary each call it a cybersecurity or cyber incident, and neither uses the word ransomware. The word comes from the leak site listings.
- Education has no sector of its own on this page, so schools and universities sit here.
Attribution: Vanderbilt, read directly; The Record; Richmond through WTVR and 12 On Your Side; Calvary through Cyber Daily alone.
The City of Osaka's leak is in the Arara item in the cross sectoral section. The Bookeasy item there covers four state and territory parks agencies.
Retail and consumer
Nitori: up to 37.62 million customer records may have leaked. [Reported]
- Nitori Holdings, a major Japanese home furnishings retailer, said on 9 November that up to 37.62 million customer records may have leaked after unauthorised access (The Star; Chunichi Shimbun).
- The records include the email and home addresses of members and online shoppers. The company says login passwords and credit card information were not compromised.
- Nitori confirmed the access on 6 November, and an investigation with an outside body found it took place in mid to late October. It has warned customers about phishing and suspicious phone calls.
- Every source counts records, not people.
Attribution: The Star and Chunichi Shimbun, each reporting the company's own announcement. Nitori's page timed out.
Iris Ohyama: details of up to about 7.7 million people, from its online shop's cloud service. [Reported]
- The Japanese homewares maker disclosed on 5 November, as dated context, that personal information of up to about 7.7 million people may have leaked from an external cloud service used by its group's online store, Iris Plaza. The story went out in this week's flash of 9 November (BleepingComputer; Business+IT).
- The figure is up to 5.8 million members and former members and 1.9 million non-members. Members' passwords were stored encrypted, and the company says card details were not stored in that service.
- It found a suspicious file on 28 October, stopped the store and told police and Japan's privacy regulator. As of 5 November it had received no reports of misuse or financial loss.
Attribution: BleepingComputer and Business+IT. The company's own notice refused the read.
Wyndham: a password inside its Android app opened a customer database. [Claimed]
- TechCrunch reports that the hotel company's Android app carried a password in its source code that opened an internal customer database hosted by Salesforce. The security researcher Robert Wiggins found it.
- TechCrunch says the database held customer names, email addresses and Wyndham member numbers, along with login tokens and company files. The password was revoked soon after TechCrunch contacted Wyndham.
- Wyndham told TechCrunch it has taken steps to address the issue and has no evidence of any unauthorised access. It would not say how long the password had been in the app, or whether customers will be told.
Attribution: a researcher's finding through TechCrunch alone, with Wyndham's reply to it, so lighter. No Wyndham notice of its own was found.
The Bookeasy, Arara and Fireworks AI items in the cross sectoral section are this sector's other priorities: each is a supplier holding a brand's customers. No material change to Who Gives A Crap's breach, in last week's Summary, was found. The Expedia claim is under circulating.
Transport and logistics
Maersk: data possibly taken from an acquired company's old system. [Confirmed]
- Maersk says it is investigating an IT security incident involving the possible theft of data from a legacy system that came from a previous acquisition. It says its findings so far strongly suggest no sensitive data has been disclosed (The Register; BleepingComputer, 11 November).
- Claimed. A group calling itself WhiteArmor claims on its leak site to hold about 100 GB from the former LF Logistics platform, which Maersk bought in 2022. It demands US$5 million in bitcoin (Ingeniøren).
- Maersk's statement, as reported, names neither the group nor the acquisition.
- A business that buys another takes on its systems, accounts and data, and the old systems are the ones nobody watches. That reading is ours.
Attribution: Maersk as reported by The Register and BleepingComputer, because its own page refused the read; the group's claim through BleepingComputer and Ingeniøren.
Uber's drivers are in the Fireworks AI item in the cross sectoral section, and the Bookeasy item there covers bookings for national parks in four states and territories. No Qantas update on the Amadeus breach after 4 November was found.
Professional services
No incident at a professional services firm in the window met the bar. Baker Tilly's notices are under circulating. This sector's first priority is the N-able item in the cross sectoral section, because managed service providers use N-central to reach their clients' computers. The AI voice fraud under financial services relied on fake lawyers at major law firms.
Circulating but not carried
- Expedia. N0n's leak site claim to hold hundreds of millions of customer records, put at 830 million loyalty rows in one account. No proof has been published, and no reply from Expedia was found.
- Baker Tilly. Notices from 9 November about client data taken through a third party platform, seen only on law firm and claims sites.
- Mirae Asset Securities, South Korea. Disclosed on 7 November, outside the window: account numbers of about 3,800 customers and details of more than 35,000 staff and contractors. The Financial Services Commission counts more than 470,000 records leaked from 37 companies in a month, as Yonhap reports it.
- Brenner Base Tunnel. BBT confirmed access to data after an Everest listing on 17 October, which is older than the window.
- Michigan State University. A gang's claim of 1.3 million students, reported by The Record alone.
- Hoag Memorial Hospital Presbyterian. On California's breach list since 4 November, but its letter would not read as text.
- Richmond Public Schools and Interlock. Interlock's claim to the attack, seen on tracker sites only.
- CSCTelecom, 125 million texts. Cybernews alone.
- NEAR AI. About 100 users affected through a flaw in the SGLang software. Small, and a single account.
- A Check Point gateway to ESXi wiper lead. The newsletter's linked research was an April paper on other ransomware. It is separate from the Check Point entry in the catalogue.
- A dental group's count of 341,039 Texans. It appeared in a newsletter and was found in no outlet.
- Stolen Entra ID app secrets used against 27 organisations. Todyl's account of 3 November, from before the window and resting on one vendor's telemetry.
- Revolve, 1.5 million. Disclosed on 29 October and not confirmed by Revolve.
- A tip that a law firm told its clients of a breach. Not stood up.
- Southwire, UT Arlington, Sportsman's Guide, Pekin Insurance, Chartboost and Infutor. Leak site listings only.
Start a conversation
Want this every week?
It goes out to the organisations we work with. Ask and you are on the list, and there is nothing else attached to it.
- Every enquiry is read by a senior leader. There is no sales sequence behind this form.
- Nothing is resold to you and no vendor introduction is waiting at the other end.
- A first conversation is a conversation, not a scoping call with a proposal attached.