Window: approximately 12 November 09:00 SGT to 13 November 09:00 SGT. New and overnight developments, deduplicated against the 12 November edition, with gaps stated plainly.
Huntress: 288 hosts across 48 organisations hit in about 90 minutes. The RMM+ extension accepted a blank access key. The technician-trap spread was designed but has not been observed to succeed.
4EverAble2Shell and Agent2Shell PoCs published 12 Nov. Exploitation claimed within hours (21 servers, miners) — contested by N-able. Disable the Legacy API.
Brazilian insurer/pension subsidiary confirms registration data copied — names and CPFs for all records. No financial or authentication data per preliminary assessment.
The Barracuda pair falls due today with no new development, so it is not carried forward as a watchlist row.
The RMM crisis deepened overnight: ScreenConnect is now a named worm, and N-able’s Hotfix 2 has already been bypassed. Huntress published the full SCKit picture on 12 Nov: 288 hosts across 48 organisations hit in about 90 minutes, spread through the RMM+ extension, which accepted a blank access key in versions 1.0.31/1.0.32 and bypasses ScreenConnect accounts, roles and MFA. The toolkit’s RelayTrap steals session cookies and is designed to trap any technician who connects — but Huntress is explicit that the technician-to-panel spread has not been observed to succeed, a correction to yesterday’s framing. Separately, researcher Sean Bales published two new N-central chains, 4EverAble2Shell and Agent2Shell, tested against Hotfix 2; Previdian and Huntback report probing within hours and 21 production servers mining XMRig, 19 of them on Hotfix 2 — while N-able says it has seen no compromise using the new flaws. The one solid discriminator: the server chain needs the Legacy API enabled, which is not the default. Disable it, and hunt for miners and for requests to /dms/services/ServerUI.
Brazil’s Caixa insurer disclosed a 7.6-million-customer breach, Korea’s and Taiwan’s regulators moved on last week’s breaches, and a GitHub credential-stealing campaign hit tens of thousands of repositories. Caixa Vida e Previdência confirmed attackers copied registration data for all 7.6 million active customers — names and CPFs throughout, plus addresses, phones, emails and birth dates in part of the records — detected 6 Nov and disclosed 12 Nov; the “no financial data” conclusion is explicitly preliminary, and some wire reporting says the data was published while the company says it has no evidence of public disclosure or misuse so far. In Korea, the Financial Services Commission will start a special audit of the breached Financial Security Institute from 16 November and the FSI president apologised; in Taiwan, the FSC gave SinoPac until the end of 12 Nov to file its full report, is examining vendor due diligence and has not ruled out a fine. In the US, Sawyer Savings Bank confirmed a breach of 16,069 people in the Storm-1175 campaign against small lenders, which Microsoft links to N-able exploitation via MSPs. And the wfx-loot campaign used stolen GitHub credentials to push a CI secret stealer into 61,764 repositories, exfiltrating credentials — including AI service keys — from at least 1,806 of them.
Citrix patched a 31-CVE batch, a public PoC now exists for the Apache mod_http2 RCE, and Check Point confirmed PHP2Shell exploitation on its own gateways — with a wiper. Citrix’s 12 Nov bulletins fix 31 CVEs (one outlet counts 12 headline flaws — the count is unresolved between sources), worst a 9.5 memory flaw and a 9.4 Console authentication bypass; no exploitation or PoC reported for the new flaws. A third-party research PoC for Apache CVE-2026-119618 was published on 12 Nov and verified as substantive research code; in-the-wild exploitation is still not confirmed. Check Point reports attackers chained PHP2Shell with a Gaia privilege-escalation flaw (CVE-2026-123039) to root on unpatched gateways, planted the TSGA backdoor, pivoted to a VMware ESXi host and ran a wiper — “a couple of dozen” systems in one region. Unit 42 separately published Cairn, a Chinese-speaking operator’s largely autonomous AI intrusions against at least 19 organisations, and Arctic Wolf tracked BlueNoroff’s AI-deepfake lures against 677 finance and crypto decision-makers. Europol’s Operation Endgame seized the Gootloader, CastleRAT and StealC infrastructure and secured 6.4 million victim datasets. No new CISA KEV additions were made on 12 November.
Group Name | Target Sub-sector / Geography | TTPs (MITRE ATT&CK where available) | Impact
| Group | Target | TTPs | Impact |
|---|---|---|---|
| Unknown actor — ScreenConnect SCKit worm — Update, previously covered (Huntress, 12 Nov) | Managed service providers and their clients; MSPs serve small banks and financial firms | No MITRE mapping published. Attackers used compromised cloud-hosted ScreenConnect panels’ “Run Command” feature to push a rogue client and the 16-file SCKit toolkit. SCKit installs scheduled task “TrapWatch” (SYSTEM), a modified client (RelayTrap) that steals the technician’s .aspxauth session cookie with a SQLite tool and tries to replay it against the panel, then uploads a payload to the panel’s Shared Toolbox to reach every session the technician can access. Huntress links the entry to the third-party RMM+ extension v1.0.31/1.0.32, which accepted a blank access key when none was configured, bypassing ScreenConnect accounts, roles and MFA; RMM+ commands appeared in logs with no corresponding logins. The RelayTrap C# code appears LLM-generated, per Huntress. | Confirmed (Huntress blog, 12 Nov 10:00 SGT; updated 12:34 SGT). At least 288 hosts across 48 organisations, all hit within about 90 minutes (about a third in the first 10). Affected builds named: 26.6.6.9747 and 26.7.6.9783. Correction to yesterday’s framing: the technician-trap spread is designed behaviour reconstructed from the toolkit’s code — Huntress states it has not observed a successful jump from an infected endpoint to a technician workstation, or from a workstation to other endpoints; the console-takeover attempt failed in every recovered host log checked. ConnectWise (12 Nov advisory) says it found no evidence of compromise of ConnectWise infrastructure or a vulnerability in ScreenConnect itself; five RMM-family extensions were disabled across ~400 partner environments as a precaution, and extension presence alone does not establish compromise. Initial access and full scope remain under investigation. New IOC: cdn007-digitaldownloads.com (C2, originally flagged by BlackPoint). New CVE: CVE-2026-127063 (see watchlist) — its “static API secret” wording differs from Huntress’s blank-key account; treat as the same extension weakness pending vendor confirmation. BlackPoint links the Relay/Trap behaviour to CVE-2026-84869 — a claim only, not confirmed. One aggregator claimed CVE-2026-84869 was added to CISA KEV; that conflicts with the verified KEV version 2026.11.10 (1,779 entries, no 12 Nov additions) and is not reported here. |
| Unknown actor — N-able 4EverAble2Shell / Agent2Shell — Update, previously covered (12 Nov) | RMM platform used by MSPs serving small banks and financial firms | No MITRE mapping published. 4EverAble2Shell: unauthenticated RCE as root on the N-central server, chaining Apache Axis CVE-2026-90436 (9.8) with a separate N-central privilege escalation (Previdian tracks as VULN-2026-0009, no CVE yet; abuses the privileged setLDFileOwner helper and a symlink to make a root-owned script writable, then invokes serverControl). Requires the Legacy API to be enabled (non-default). Agent2Shell: chains 10+ flaws to run commands as SYSTEM on managed Windows agents. PoCs published 12 Nov by Sean Bales (repo now under the renamed account MSPNightmare/CVEs), tested against Hotfix 2 (2026.4.1.26). | Exploitation claimed within hours — contested. Previdian saw probing from 157.20.241.16 from ~08:30 UTC (16:30 SGT), with full-chain attempts after; Huntback counted 21 production N-central servers mining XMRig at 12:50 UTC (20:50 SGT), 19 of them on the Hotfix 2 build, across the US, Canada, UK, Germany, France, Sweden and South Africa; the miner uses about half the CPU cores and the payload is pulled by curl from 157.20.241.16. Huntback identified victims via mining-pool identifiers (nc_ + server IP). N-able opened incident 202609 on 12 Nov, says hosted N-central is not affected, has restricted some features/integrations, switched off the Intune integration as a precaution, and says it has seen no signs of compromise using the new flaws. Huntress had not observed exploitation in partner environments as of its 12 Nov 19:30 SGT update. The two accounts conflict — flag, do not resolve. No fix existed for the new chains at publication; a follow-up hotfix is planned. N-able is not advising customers to take N-central offline, and says servers on HF2 with Legacy API disabled are not exposed. Previdian warns the public PoC is unverified and may contain malware. Hunt markers: requests to /dms/services/ServerUI referencing jline.internal.TerminalLineSettings; unexpected xvd.service systemd services, mining processes, files under /tmp/.xvd or /var/tmp/.xvd. Any server showing these IOCs should be treated as fully compromised. |
| Unknown actor — Caixa Vida e Previdência — NEW (confirmed, 12 Nov) | Life-insurance and private-pension subsidiary of Caixa Seguridade / Caixa Econômica Federal, Brazil; 7.6 million active customers | No MITRE mapping published. Attacker and entry route not disclosed; investigation ongoing. | Confirmed (company disclosure via Caixa Seguridade market notice, 12 Nov; Brazilian press from ~02:00 SGT 13 Nov). Unauthorised access to company systems; registration data copied outside the company’s environment. Affected base covers all 7.6 million active customers plus an undisclosed number of former customers — the total number of people affected was not disclosed. Data exposed: names and CPFs (Brazilian tax IDs) for all records, plus addresses, phone numbers, emails and dates of birth in part of the records. The company says, on a preliminary assessment, there is no indication of compromise of financial or authentication data (policy values, pension balances, bank data, passwords), and no evidence so far of public disclosure or misuse. Operations continue normally. Detected 6 Nov; reported to ANPD (data protection authority) and other authorities 11 Nov; disclosed 12 Nov — a six-day gap the Brazilian press has flagged and the company has not explained. Conflict to flag: some wire reporting (ANSA / La Nación) says the data was published and rounds the figure to “circa 8 million”; the company says it has no evidence of public disclosure or misuse so far. Do not present either as settled. This is the insurer/pension subsidiary, not Caixa Econômica Federal’s current accounts; the company states other subsidiaries (Caixa Cartões, Caixa Loterias) and the bank’s accounts were not affected. Federal Police involvement is unconfirmed; ANPD is verifying the company’s transparency and mitigation measures. |
| CL-STA-1238 — Cairn framework — NEW (Unit 42, 12 Nov) | At least 19 organisations, Aug–Sep 2026: 7 in the semiconductor / advanced-manufacturing supply chain, 3 in defence / defence policy, 4 in human rights / civil society | No MITRE mapping published. Chinese-speaking operator used the open-source AI pentest framework Cairn to run 25 largely autonomous operations, supplying only a target domain and a one-line goal; Cairn planned, executed and logged results itself via a shared-blackboard fact/intent loop, with a custom “AIScan Runner Skills” pack and a Chinese authorisation pretext neither model refused. Models used: Zhipu GLM 5.2 (13 ops), DeepSeek V4 Flash/Pro (8 + 4). The exploited weaknesses were all known: auth bypasses, IDOR, mass-assignment privilege escalation, weak/default credentials, exposed cloud APIs, dated CVEs. Individual operations ran 20 minutes to 50+ hours. | Confirmed (Unit 42 research, published 12 Nov 23:00 SGT). Unauthorised access or data retrieval was confirmed at 10 organisations. Outcomes included >1 million member records plus ~31,000 CRM contacts and staff password hashes at a US teachers’ union; military and government identities; persistent admin accounts and an API token created in a production system; and a live industry-standards ballot document overwritten via unauthenticated cloud-storage write. Unit 42 assesses with moderate confidence this is China-nexus intelligence collection, likely one operator or a small team hitting 19 organisations in two weeks. BFSI relevance indirect but material: autonomous AI intrusion at scale against internet-facing apps, using exactly the identity and configuration weaknesses banks present; semiconductor supply-chain targeting also touches financial-sector hardware supply. Unit 42’s key point: autonomy raised volume and coverage, not complexity — identity hygiene, attack-surface reduction and patching still work. IOCs: 154.44.31.4 (Hong Kong, Cairn controller), 82.29.129.54 (US egress/staging node). |
| BlueNoroff — NEW (Arctic Wolf Labs, 12 Nov) | Crypto and venture-capital decision-makers; targets are founders and C-suite with signing authority | No MITRE mapping published in the blog text — the full report PDF includes a MITRE mapping, which was not fetched; do not invent IDs from the blog alone. Lures evolved from fake Zoom/Teams meetings to shared financial documents, fake Chrome PDF-viewer updates, fake media players (a fabricated “NovaStream” player), identity-verification prompts, hardware-wallet access and security-audit workflows. AI is now a production multiplier: deepfakes built with Google Generative AI, Grok Imagine, Wondershare AIGC and GPT-4o, no longer distinguishable by visual inspection (Arctic Wolf used provenance evidence). The intrusion mechanism itself is unchanged — it still depends on the victim running a “fix”. | Confirmed (Arctic Wolf Labs, 12 Nov, based on access to the actor’s own infrastructure and media library). The actor’s target list grew from ~400 people / 20 countries (April) to 677 individuals across 58 countries, 84% founders / C-suite. 98 high-confidence domains documented, 73 previously unreported. BlueNoroff is a financially motivated Lazarus subgroup (DPRK RGB). BFSI relevance high: the targets are finance and crypto decision-makers, and AI-deepfake social engineering is the AI-as-weapon item of the day. Directly related: the FBI’s FaceStealer announcement below (same DPRK crypto-theft ecosystem, different cluster). |
| Sapphire Sleet / UNC1069 — “FaceStealer” — NEW (FBI, 12 Nov) | Crypto / Web3 workers and open-source software maintainers; their contacts | No MITRE mapping published. Actors lure victims onto video chats on imposter Zoom / Teams / Meet, secretly record webcam video, then use the footage to target the victims’ contacts. The FBI says actors are starting to use AI to create deepfake videos of well-known crypto/blockchain figures. | Confirmed (FBI announcement, 12 Nov; Reuters 13 Nov 05:40 SGT). Over $400 million in cryptocurrency stolen in the past 18 months. FBI/DOJ disrupted cloud infrastructure used in the scheme and identified/seized 900+ victim webcam videos. Attributed to Sapphire Sleet / UNC1069 (also Stardust Chollima, APT38, BlueNoroff). This cluster is separate from the group behind the $6.3bn Binance hack. Separately, DOJ froze >$1.2bn in digital assets tied to North Korean actors and the Binance theft — keep distinct from FaceStealer. The underlying IC3 FLASH advisory (CSA-261112) is dated 11 Nov; the public announcement and disruption are the in-window event. BFSI relevance: direct for crypto exchanges and digital-asset firms and their staff; deepfake video-call social engineering is a live fraud vector for finance; targeting of open-source maintainers is a software-supply-chain risk. Also on 12 Nov the FBI announced Operation TorNado (354 dark-web drug arrests, $234m seized) — law-enforcement news, low BFSI relevance. |
| Unknown actor — Korean Financial Security Institute — Update, previously covered | Financial-sector security assessment body, South Korea; assessment staff at banks, insurers, securities firms | No MITRE mapping published. Attacker created a fake account impersonating a Financial Supervisory Service (FSS) official, exploiting a weakness in the authority-approval process to gain FSS-level access and view/download submitted material. No actual FSS account was taken over and no FSS account data leaked, per FSI. | New in-window (12 Nov, Korean press, translated): the Financial Services Commission will start a special audit (특정감사) of FSI from 16 November, with experts and related agencies, following the on-site inspection that began 11 Nov (Yonhap, 12 Nov 16:09 SGT; confirmed in Dailian, 13 Nov 08:47 SGT). FSI president Park Sang-won publicly apologised at FISCON 2026 on the morning of 12 Nov (VIP event cancelled). FSI has asked police to trace the 236 IPs; the Seoul Metropolitan Police Agency cyber unit has begun investigating. Scope unchanged: 48 mutual-finance firms, about 130 files (about 200 administration forms, about 300 personal records, about 350 system-user records). All 48 firms are mutual-finance units (regional agricultural cooperatives, credit unions, forestry cooperatives, Saemaul Geumgo); no large firm among them. FSI says the breached system does not process customer financial transactions or collect vulnerability-assessment results. The 2022 FSC special-audit warning about the same system, surfaced yesterday, remains a live question, not a finding. |
| Unknown actor — Bank SinoPac — Update, previously covered | Bank, Taiwan; 418 customers hit in the 3 Nov incident | No MITRE mapping published. External attackers exploited a vendor-supplied “middle platform” (中台, middleware layer of the bank’s new online-banking app) security setting; 427 abnormal transactions, NT$41.76 million siphoned to a Taipei Fubon Bank account; the bank reimbursed customers on 3 Nov. | New in-window (12 Nov, Taiwanese press, translated): the FSC required SinoPac’s full written report by the end of 12 Nov; said it will examine whether SinoPac failed vendor due diligence (banks must vet suppliers for malicious code/backdoors) and may fine it; convened a CISO meeting of financial institutions — figure conflict: 67 CISOs per CNA vs 76 institutions per UDN — flag, do not resolve — with four defence priorities (reduce attack surface, dynamically prioritise high-risk patching, more security resourcing, careful adoption of defensive AI) and noted it has issued an agentic-AI security advisory for the sector; and told SinoPac to share the incident via F-ISAC. The disputed system is used by no other Taiwanese bank. Separately, social-media reports of isolated unauthorised transfers at Hua Nan Bank (1 person, 2 × NT$100,000) and Taipei Fubon Bank (1 person, 2 × NT$10,000) were addressed by the FSC, which says those cases differ from SinoPac’s (device binding, password + OTP + biometrics used; one customer has reported to police). Context in the same window: Taiwan’s Ministry of Digital Affairs said 12 Nov that 10+ companies recently suffered breaches totalling ~88 million records, and Taiwan faces ~2.8 million cyberattacks a day. |
| Storm-1175 — Sawyer Savings Bank — NEW (confirmed, 12 Nov) | Small US lenders; Sawyer Savings Bank (Saugerties, New York; ~$285m assets) and Insight Credit Union | No MITRE mapping published. The compromise was inside the bank’s managed service provider’s environment, not the bank’s own systems. Microsoft tracks the leak site as Storm-1175, which it says is likely exploiting a flaw in N-able N-central software used by MSPs, and began deploying a new ransomware strain (“StormEncryptor”) on 2 Aug. No specific CVE is named in the filings. | Confirmed (bank breach notification + AG filings; American Banker 11 Nov, updated 12 Nov 05:00 SGT; TMC Insight 12 Nov 17:48 SGT). Sawyer Savings Bank notified 16,069 people that names and Social Security numbers were taken, per the Indiana AG breach report; the bank closed all branches for a week in August. The bank says there is “NO evidence that debit card numbers, financial account numbers” or core banking system data were involved. Incident identified 3 Aug 2026; file review completed 21 Sep; notification letters mailed 9 Oct. The bank’s filings do not name the MSP or say whether ransomware was involved. Storm listed Sawyer on its leak site on 7 Aug. Wider campaign: Storm-1175 has listed five US banks holding ~$6.1bn in assets plus Insight Credit Union (~$6.7bn combined); only Sawyer and Insight Credit Union have confirmed breaches — the other four banks had not publicly confirmed or denied as of publication. Insight Credit Union reported a breach in Texas and Vermont AG filings dated 6 Nov 2026; data may include names, SSNs, driver’s licence / government ID numbers, financial account numbers and card information. Key BFSI takeaway: third-party / MSP concentration risk for small lenders. Flag: the 16,069 total comes from the Indiana AG’s published breach list, not from the bank’s own per-state filings (72 MA / 8 NH / 25 VT residents plus 6 RI in the sample letter). |
| Payoutsking — iLending & J.C. Flowers & Co. — Catch-up / backfill (first seen 3 Nov) | US financial services (auto-loan refinancing / investment firm) | No MITRE mapping published. Payoutsking is assessed by Zscaler ThreatLabz as former BlackBasta affiliates; typical access is Teams vishing + Quick Assist / AnyDesk / TeamViewer, not CVE exploitation. | Reported / unconfirmed — leak-site claim only. Payoutsking’s leak site names “iLending & J.C. Flowers & Co.” (domain save.ilendingcarloanrefinancing.com), with the standard threat that the full leak will be published unless a company representative makes contact. First seen/posted 3 Nov 2026 (ransomware.live, ransomlook.io, AcidPeak); aggregators re-dated it 12 Nov when they re-indexed it — the 12 Nov date is a re-publication artefact, not a new listing. Three independent trackers list the same post, which evidences the posting itself — but nothing about whether a breach happened, what data was taken, or the claimed volume. Neither named organisation has confirmed anything. Flag: it is not established that iLending and J.C. Flowers & Co. are a single combined entity — the listing pairs two names and the relationship is unverified. Included here as a missed backfill, labelled plainly, not as a 12 Nov incident. |
| Unknown actor — Banco do Brasil — NEW (reported / unconfirmed) | Largest bank in Brazil by assets; retail customers | No MITRE mapping published. No intrusion detail published. | Unconfirmed — threat-actor claim only (Undercode News, 12 Nov, citing Dark Web Intelligence). An actor claims to offer 22,705,600 Banco do Brasil customer records (~50% of accounts), with samples allegedly containing customer IDs, CPF numbers, names, dates of birth, ID documents and account info, in MongoDB format. The actor claims extraction stopped after the bank detected suspicious activity. No independent confirmation; no evidence provided beyond the claimed samples. Do not present as a confirmed breach. Watch item only — and do not confuse with the confirmed Caixa Vida e Previdência breach above, a different institution. |
| INC_RANSOM — Shenwan Hongyuan (H.K.) — Update, previously covered | Hong Kong brokerage (HKEX: 0218) | No MITRE mapping published. No intrusion detail published. | New in-window development — actor-claimed haul detail only, still reported / unconfirmed. Trackers reproducing the INC_RANSOM leak-site listing (12 Nov) claim 21,002,068 files / 9.97 TB exfiltrated, including eight months of FICC proprietary trading positions (~$197m YTD revenue), a derivatives portfolio with HKD 2.8bn net exposure, ISDA master agreements with 20 global investment banks, 22 quarters of PwC audit packages, SFC 2024 inspection materials, 162 resigned-employee mailbox archives, full network/infrastructure credentials, 796 NTLM hashes, 509 plaintext credentials and 896 employee records, plus a claim of no cyber insurance. Undercode News’ 12 Nov assessment explicitly states the listing alone does not establish data theft, no sample or deadline was published, and the company disclosed only a suspected extortion message on 23 Oct, reporting to Hong Kong Police and stating trading/settlement systems normal with no customer asset loss. Do not present the 9.97 TB figure as confirmed. |
| Unknown actor — The Atlantic Federal Credit Union — NEW (reported) | Credit union, US | No MITRE mapping published. No intrusion detail published. | Reported — thin sourcing. A breach notice filed with the Vermont AG, published 12 Nov, reportedly exposes SSNs and financial account information; breach date and affected count not stated. Source is a class-action aggregator only; treat as reported, low detail, and watch for the bank’s own notice or AG filing. |
| DarkLunnen — Catch-up (Group-IB; publish date unconfirmed, appears ~10 Nov) | Government, defence and aviation targets in Europe; no financial victim named | No MITRE mapping published. Unclassified Russian-speaking operator; exposed Latvia server 46.8.236.121:8090 exposed /root/ (35 of 457 files recovered, 3.1 GB). Used a modified watchTowr FortiGate CVE-2024-55591 PoC creating backdoor super-admin “service”; ~600 vulnerable FortiGates in Turkey, 16 Ukraine, 14 Azerbaijan, 1 Algeria; a stolen Romanian FortiGate config yielded 10 cleartext VPN passwords; 5,775 brute-forced VNC endpoints; 12,183 Hikvision camera IPs scanned (CVE-2017-7921), 8 vulnerable; Ukrainian defence/aviation targeting including git-dumper source-code theft attempts; Exchange webshells; a possible German network compromise with a stolen NTLM hash; Sliver C2; Moonshot AI’s Kimi coding assistant installed (use unestablished). | Reported (Group-IB blog) — catch-up, not 12–13 Nov news. The publish date could not be confirmed from the page text; embedded image filenames are dated 10 Nov 2026, suggesting publication around then. Not in the 11 or 12 Nov editions. BFSI relevance indirect — the FortiGate / Exchange / VPN-credential theft pattern is exactly the perimeter-to-credential path banks face; no financial victim is named. Included plainly as a catch-up with the date caveat. |
| Unknown actor — AU Small Finance Bank — Update, previously covered | Small finance bank, India; vehicle-loan customers | No MITRE mapping published. No intrusion detail published. | New in-window (RDU Global, 12 Nov 10:30 SGT): the bank is investigating a possible breach of limited customer data linked to its vehicle-loan portfolio, hosted on a third-party platform operated by Salesforce. Core banking systems not indicated as compromised. The bank has not detailed the data involved. Reported / unconfirmed — bank investigating. Pairs thematically with the Klue / Salesforce OAuth incident in Supply Chain below. |
| Unknown actor — Seven Bank — NEW to this brief (disclosed 11 Nov; Japanese press coverage 12 Nov) | Bank, Japan; foreign resident account holders | No MITRE mapping published. Cause and intrusion route not disclosed; investigation ongoing. | Confirmed (bank disclosure, 11 Nov; Japanese press 12 Nov). Seven Bank said its residence-card submission site (for foreign resident account holders) and account-cancellation reception site were accessed without authorisation; traces found 9 Nov, both sites suspended. Up to ~150 residence-card face-image records held on the submission site may have leaked — the figure is a maximum, not a confirmed count. Affected: people notified of document deficiencies, or who used the site 16 Oct–9 Nov 2026. No unauthorised remittances, account cancellations or data tampering confirmed; ATMs, internet banking and debit services unaffected; no misuse confirmed. Small scale, but identity-document data is fraud-usable. |
| Unknown actor — SMBC Nikko Securities — No new development in-window | Securities firm, Japan; members of employee stockholding associations at listed companies | No MITRE mapping published. | No change in-window. Figures unchanged: up to 56,487 members; names, addresses, holdings, balances, dividend/tax data, transfer-destination account info; no phone numbers or emails (never stored). No secondary harm confirmed as of 11 Nov. The 12 Nov FNN/Tetoran coverage adds no new facts. |
| Unknown actor — SMBC Consumer Finance — No new development in-window | Consumer finance, Japan; customers of partner banks, credit unions and online banks | No MITRE mapping published. | No change in-window. Total remains up to 320,118 as announced 5 Nov. |
| Unknown actor — Mirae Asset Securities — No new development in-window | Securities brokerage, South Korea | No MITRE mapping published. | No change in-window. FSS on-site inspection began 9 Nov (previously covered). Figures unchanged: ~3,800 customer account numbers; 35,255 staff personal data. |
| Unknown actor — Philippine SEC — No new development in-window | Securities regulator, Philippines | No MITRE mapping published. | No change in-window. Facts unchanged: detected 6 Nov, notified NPC 9 Nov; names, usernames, emails, mobiles, birth dates, genders, addresses and TINs may have been exposed; no passwords leaked, no data altered, no company records affected (per SEC). |
| Unknown actor — Crédit Mutuel Arkéa — No new development in-window | Cooperative banking group, France | No MITRE mapping published. | No change in-window. |
| Marlanwg — Bank Rakyat — No new development in-window | Cooperative bank, Malaysia | No MITRE mapping published. | No change in-window. |
| Storm-3358 — No new development in-window | Financial sector, Southeast Asia | No MITRE mapping published. | No change in-window beyond the Unit 42 Cairn post above, which is vendor analysis of autonomous AI tooling, not a new Storm-3358 campaign, victim or tool development. |
| Deadlock — Fidex, Coopebis, Kadick — No new development in-window | Insurance broker (Mexico); savings and credit cooperative; agency banking (Nigeria) | No MITRE mapping published. | No change in-window. All three remain leak-site claims only. |
| N0n / Krybit — No new development in-window | Wealth management, Canada; payment-card network, US; conglomerate with insurance interests, Australia | No MITRE mapping published. | No change in-window. The 9 November listings remain leak-site claims only, no victim named, no data released. |
| TraderTraitor — Binance — No new development in-window | Crypto exchange, global | No new MITRE list in-window. | No change in-window. No Recovery Update #10 has been published. |
| Aoqin — Tyfone — No new development in-window | Digital/mobile banking platform provider, US; its customers are credit unions and community banks | No MITRE mapping published. | No change in-window. The leak-site claim (35 GB, per Ransomware.live/QPulse; Undercode disputes the volume) remains reported / unconfirmed; Tyfone has not confirmed an incident. |
| INC_RANSOM — Raymond James Financial — No new development in-window | Diversified US financial-services firm | No MITRE mapping published. | No change in-window. The leak-site listing, discovered 12 Nov 10:04 UTC, remains reported / unconfirmed; no data volume, sample or company confirmation. Do not conflate with the separate January 2026 Massachusetts AG filing or the July 2026 Werth Wealth Management incident. |
No new in-window item was found for Insight Credit Union beyond the Storm-1175 confirmation above, Vanguard Group, Gopax, Mercado Bitcoin, NTT Docomo, Pekin Insurance, Higginbotham, Qilin / AXE – EXPERTS, the two unnamed Canadian firms (8 Nov), Scry Analytics, FIN7 / Carbanak, Scattered Spider, RansomHub, Cl0p, or a FI-specific Initial Access Broker. Absence of reporting is not evidence of absence of activity.
CVE ID | CVSS Score | Affected Product | Exploitation Status | Relevance to Finance
No new CISA KEV additions were made on 12 November. The catalogue stands at 1,779 entries (catalogue version 2026.11.10). The Barracuda pair (CVE-2026-116606 / CVE-2026-111560) falls due today, 13 November, with no new attribution, IOCs or PoC in-window — per the standing rule a due date alone is not a new development, so it is not carried forward as a watchlist row. The four 9 November KEV additions fell due 12 November, likewise with no new development, and are not carried. The other 10 November KEV additions (Outlook CVE-2026-124194, Chrome CVE-2026-125350, Cisco BroadWorks CVE-2026-20320, Openwave CVE-2026-111793) likewise have no new development in-window and are not carried. Verification limit: the conclusion rests on the catalogue version and entry count plus the absence of any 12 November-dated addition reporting across multiple sources — not a live catalogue read.
| CVE | CVSS | Product | Status | Finance relevance |
|---|---|---|---|---|
| Citrix NetScaler November batch — 31 CVEs NEW 12 NOV | 9.5 (highest) | NetScaler ADC, NetScaler Gateway, NetScaler Console, Secure Access Client (Windows/Linux) | Not known exploited — no PoC reported | New in-window (Citrix bulletins, 12 Nov). Four bulletins fix 31 CVEs. Headline flaws: CVE-2026-107412 (9.5, memory flaw, highest of the batch); CVE-2026-120868 (9.3, memory overflow in packet processing, affects all ADC/Gateway deployments); CVE-2026-120869 (improper authentication in NetScaler Console, unauthenticated bypass of authentication to access restricted resources — attack is adjacent-network, needs Console IP access); CVE-2026-118268 (8.5, Secure Access Client for Windows privilege escalation to SYSTEM) and CVE-2026-118269 (Linux equivalent). Fixes: ADC/Gateway 14.1-73.57 and 13.1-64.39; Console 14.1-74.34. Citrix says it is not aware of any exploits for the announced vulnerabilities; the fixes followed a deeper codebase review using AI-assisted code analysis. Conflicts — flag, do not smooth over: The Hacker News and others say “12 flaws” while SecurityOnline says four bulletins fixing 31 CVEs (28 memory flaws in ADC/Gateway alone) — the 12-flaw figure appears to count only the headline set; the true CVE count is unresolved between sources. CVSS for CVE-2026-120869 is 9.4 per THN and 9.3 per cve.now / StackFlag. NetScaler ADC/Gateway is widely deployed as a remote-access/ADC front-end in financial firms; the Console bypass is the priority patch (management-plane exposure). Separately, CitrixBleed 2 (CVE-2025-5777) is still actively exploited — OffSeq honeypots recorded 314,119 exploitation attempts from 22 source IPs in the last 30 days, last seen 12 Nov 22:40 UTC (13 Nov 06:40 SGT), with DragonForce ransomware as the severe outcome in past cases. |
| CVE-2026-119618 STATUS — PoC NOW PUBLIC | Critical — no CVSS yet | Apache HTTP Server 2.4.55–2.4.70; standalone mod_h2 2.0.8–2.0.42 | Public PoC — in-the-wild exploitation not confirmed | New in-window development: a third-party research PoC was published 12 Nov (GitHub googlesky/CVE-2026-119618, created 12 Nov 13:05 SGT) — verified as substantive research code (root-cause analysis of the assign_headers() race, containerised lab, stdlib HTTP/2 PoC, ASan/TSan evidence), not a lure; the author “googlesky” is an unknown account, not one of the original discoverers, so this is a third-party reproduction and the code was not run. Conflict — flag: Wiz (11 Nov) said no exploit was available; Red Hat reportedly says a reliable exploit is public. Treat exploitation status as “public PoC now exists; in-the-wild exploitation not confirmed”. Race condition in mod_http2 assign_headers() — concurrent requests corrupt shared heap data, leading to remote code execution. Fixed in httpd 2.4.71; disabling HTTP/2 is a mitigation. Apache httpd fronts many banking web apps, APIs and customer portals; unauthenticated RCE on a public web server is a direct initial-access path. NVD lists the CVE as awaiting analysis. |
| 4EverAble2Shell / Agent2Shell — CVE-2026-90436 + VULN-2026-0009 NEW 12 NOV | 9.8 (CVE-2026-90436) | N-able N-central (on-premises servers with Legacy API enabled) | Exploitation claimed — contested (see actors) | New in-window (PoCs published 12 Nov). 4EverAble2Shell chains Apache Axis CVE-2026-90436 (9.8) with a separate N-central privilege escalation (Previdian VULN-2026-0009, no CVE yet) to unauthenticated root on the server; it requires the Legacy API to be enabled (non-default). Agent2Shell chains 10+ flaws to SYSTEM on managed Windows agents. PoC repo: MSPNightmare/CVEs — N-Central N-Able (GitHub) — genuine researcher disclosure (Sean Bales), with Previdian’s caveat that the public PoC is unverified and may contain malware. No fix existed at publication; follow-up hotfix planned. RMM used by MSPs serving small banks; a Hotfix-2 bypass with claimed live exploitation is a live patching and hunting priority. Disable the Legacy API; hunt the IOCs in Indicators. |
| CVE-2026-127063 NEW 12 NOV | 9.3 | DataComm RMM+ extension for ConnectWise ScreenConnect | Not known exploited — no PoC reported | New in-window (CVE published 12 Nov, per cve.now / ENISA EUVD). “The third-party RMM+ extension for ConnectWise ScreenConnect uses a static API secret by default.” Caveat: the CVE wording (“static API secret”) differs from Huntress’s account of a blank access key accepted when none was configured — treat as the same extension weakness pending vendor confirmation; no fix or version detail in sources read. This is the extension weakness behind the SCKit worm (see Threat Actor Activity). Check RMM+ versions 1.0.31/1.0.32 and remove or re-key the extension. |
| CVE-2026-123039 NEW 12 NOV | 7.8 | Check Point Security Gateways (Gaia DynamicCLI) | In the Wild — chained with PHP2Shell | New in-window (Check Point, 12 Nov). Local privilege escalation in Gaia DynamicCLI, chained with PHP2Shell (CVE-2026-109618/109619) in the Mobile Access Portal to gain root on unpatched Security Gateways. Both fixes are in the November Jumbo Hotfix (which also carries Check Point’s HTTP2Shell / Node2Shell fixes). The PHP2Shell fix had been available since 23 Oct — the one-week patch lag is the lesson. See the PHP2Shell row below for the campaign detail. |
| CVE-2026-121529 (“Fortinet2Shell”) CATCH-UP — CVE 10 NOV, EXPLOIT 11 NOV | Critical (no CVSS figure in sources read) | FortiClientMac 7.4.0–7.4.7; FortiClientWindows 7.4.0–7.4.7 and 7.2.0–7.2.14 | Public exploit — in-the-wild exploitation not reported | Catch-up: CVE published 10 Nov; public exploit on GitHub 11 Nov (googlesky/CVE-2026-121529 / googlesky/Fortinet2Shell) — missed by the 11 and 12 Nov editions; included here plainly as a catch-up, not as 12–13 Nov news. Origin-validation flaw, unauthenticated remote code execution on FortiClient endpoints. PSIRT FG-IR-26-194. Not in CISA KEV. FortiClient is the endpoint agent on bank staff machines; an unauthenticated RCE with a public exploit is a patching priority. Note the same GitHub account (“googlesky”) published both this and the Apache PoC above — a prolific third-party reproducer, not the original discoverer in either case. |
| Able2Shell — seven CVEs STATUS — NEW CHAINS 12 NOV | 9.3 (highest) | N-able N-central | PoC public — exploitation of new chains claimed, contested | Update in-window — see the 4EverAble2Shell row above. Original Able2Shell (9 Nov): researcher Sean Bales published PoC code for 33 alleged N-central flaws, including an unauthenticated chain to RCE/root (XXE in service import → session theft → password reset → command execution as the nable user). Hotfix N-central 2026.4 HF2 (build 2026.4.1.26), released 10 Nov, fixes seven CVEs: CVE-2026-125883 (XXE, 9.3), CVE-2026-125878 (OGNL injection), CVE-2026-125879, CVE-2026-125880, CVE-2026-125881 (SSRF), CVE-2026-125882 (RCE via path traversal), CVE-2026-91000. All hosted servers upgraded by 11 Nov 01:04 UTC. The Able2Shell PoC repository (Oracle-Security/CVEs, path “0-Days/N-Central N-Able”, 26 Python PoC scripts) is a genuine researcher disclosure — linkable as the PoC source, with the caveat that N-able has not validated all 33 findings. RMM used by MSPs serving small banks; an unauthenticated RCE with public PoC plus claimed live exploitation of the follow-on chains is a live patching priority. |
| CVE-2026-109618 / 109619 (PHP2Shell) STATUS — GATEWAY EXPLOITATION CONFIRMED 12 NOV | 9.2 | PHP; PAN-OS and other network/security products; Check Point Security Gateways | In the Wild | New in-window development (Check Point Research / advisory, 12 Nov 21:58 SGT). Attackers chained PHP2Shell in the Mobile Access Portal with CVE-2026-123039 (Gaia DynamicCLI local privilege escalation, 7.8 — see row above) to gain root on unpatched Check Point Security Gateways. They planted PHP webshells in login templates (zz*.tpl, hardcoded key ‘zz2026’), the persistent TSGA backdoor (/tmp/CPconfigdaemon, “slot_CPconfigdaemon”) with an embedded TSProxy component (/lib/libcpfwcommon.so), used the gogo scanner for internal discovery, then pivoted over SSH to a VMware ESXi host and ran the destroy_v6 wiper, leaving DREAM_ANNEXATION.txt notes. Two waves: 30 Oct against Israeli organisations (destructive activity in some) and 2 Nov across multiple countries/sectors (no destructive activity). Activity ran 30 Oct–4 Nov; “a couple of dozen” unpatched systems in one region compromised. TSGA/TSProxy are modified variants of PhiliKit (ESET previously linked PhiliKit to China-nexus UNC5221), but Check Point explicitly says the evidence does not allow high-confidence attribution — the actor may have run the whole campaign, acted as an initial-access broker, or run a false-flag operation. CSA 3 Nov: PAN-OS management interfaces exploited, chained with CVE-2026-46300 to root. Keep management interfaces off the internet; apply the November Jumbo Hotfix. IOCs: 130.94.19.204, 130.94.19.168; upload_progress_pwn string in PHP session files; process masquerading as CPconfigdaemon; scanner/wiper files under /tmp or /temp; DREAM_ANNEXATION.txt notes. |
| CVE-2026-123864 | High — no CVSS published | Node.js 22.x / 24.x / 26.x (HTTPS servers) | Not known exploited — no PoC reported | No change in-window. HTTPS heap-memory disclosure: a zero-length TLS write re-enters the HTTP parser, and the server copies other connections’ heap memory — including TLS private keys and other users’ requests — into a request body. Fixed in 26.13.1 / 24.22.3 / 22.23.5. Node.js published no CVSS scores and lists no workarounds — upgrade only. |
| Node.js batch — six further High | High — no CVSS published | Node.js 22.x / 24.x / 26.x | Not known exploited — no PoC reported | No change in-window. CVE-2026-123863, CVE-2026-123865, CVE-2026-123866, CVE-2026-123872, CVE-2026-123874, CVE-2026-126051 (HTTP/2 and TLS memory disclosure / DoS / use-after-free). Six Medium and three Low complete the 16-CVE release. Node.js states 16 CVEs; one aggregator lists 17 IDs — report Node.js’s count. |
| CVE-2026-126263 | 9.8 | FortiMail 8.0.0–8.0.1, 7.6.0–7.6.4, 7.4.0–7.4.6 | Not known exploited — no PoC reported | No change in-window. SQL injection (CWE-89): an authenticated attacker can execute unauthorised code/commands via crafted requests. Fixed in 8.0.2 / 7.6.7 / 7.4.7; FortiMail Cloud already remediated. Internally discovered with the Frontier AI model Claude Mythos. |
| CVE-2026-126155 | 9.1 | FortiMail 8.0.0–8.0.1, 7.6.0–7.6.6, 7.4.0–7.4.8, 7.2 all versions | Not known exploited — no PoC reported | No change in-window. Incorrect authorisation (CWE-863): an unauthenticated remote attacker can obtain an authenticated IBE webmail session as any IBE recipient and read or send their encrypted mail. Fixed in 8.0.2 / 7.6.7 / 7.4.9; 7.2 must migrate to a fixed release; FortiMail Cloud remediated 5 Oct. |
| CVE-2026-121653 | 9.8 | Ivanti Endpoint Manager Core, 2024 SU7 and earlier | Not known exploited — no PoC reported | No change in-window. SQL injection in EPM Core allows a remote unauthenticated attacker to achieve remote code execution. Fix: EPM 2024 SU8. Sibling critical: CVE-2026-121683 (9.4, missing authentication in Cloud Services Application 5.0.8 only; fix 5.0.8.1). Ivanti says it is not aware of any customers being exploited, and none of the 13 CVEs in its 10 Nov batch is in CISA KEV. |
| VSA-2026-062 (no CVE ID yet) | — no CVSS assigned | Xen Orchestra / Xen Orchestra Appliance | Actively exploited | No change in-window. Empty-object login token gives Admin privilege; full XCP-ng host takeover as root. Fixed in Xen Orchestra 6.9.1 / 6.10.0. IOCs: 91.211.251.112:8001 and Tor domain ustar3rqxgnskzgnwkgpofgtvxevm6cx3lxdbwt4mndtveqyaohid.onion:3333. |
| SAP November Patch Day — four CVSS 10.0 | 10.0 | SAP NetWeaver AS ABAP / Web Dispatcher / Gateway | Not known exploited | No change in-window. CVE-2026-76964 (ICM memory corruption, unauthenticated RCE — “ICM2SHELL”), CVE-2026-89126 (Web Dispatcher memory corruption, unauthenticated RCE), CVE-2026-89103 (Gateway broken access control), CVE-2026-120017 (SAML2 missing authentication). Patch internet-facing systems first. |
| Fortinet batch — eight criticals | 9.8–9.0 | FortiOS / FortiManager / FortiAnalyzer / FortiSASE | Not known exploited | No change in-window. Worst CVE-2026-126170 (9.8, JSON RPC API). Fixed in FortiOS 7.2.14/7.4.13/7.6.8/8.0.2 and FortiManager/FortiAnalyzer 7.2.13/7.4.12/7.6.8/8.0.2 — some fixes still “upcoming”; apply the FortiGuard virtual patch, keep OFTP/FGFM ports off the internet. |
| Microsoft November Patch Tuesday | Up to 9.9 | Microsoft enterprise stack (Exchange, SharePoint, SQL Server, AD, LDAP, DNS) | One exploited (Outlook CVE-2026-124194, not carried — see note above) | No change in-window. 1,094 newly released CVEs, 64 Critical. Notable criticals, none reported exploited: SharePoint CVE-2026-120002 (9.9); SQL Server CVE-2026-116945 (9.8); Windows Search CVE-2026-116928 (9.8, zero-click); LDAP CVE-2026-119958 (9.8, zero-click, unauthenticated); Windows DNS CVE-2026-88218 (9.8); iSCSI CVE-2026-88234 (9.8); Exchange CVE-2026-118207 (8.8); AD DS CVE-2026-118146 (8.8); Schannel CVE-2026-119952 (8.1). Tenable counts 1,093 CVEs in the same Patch Tuesday (one fewer than Microsoft’s 1,094) — a counting difference between trackers, not a new item. Tenable also notes the exploited Outlook CVE was discovered by a researcher using Claude. |
| CVE-2026-88771 | 9.5 / 9.8* | Citrix NetScaler ADC and NetScaler Gateway | In the Wild (KEV) | No change in-window. Public Metasploit module (6.5.9, 6 Nov). Unauthenticated pre-auth command injection as root via attacker-controlled text written to appliance logs, later executed by maintenance script ns_monuploadd_err.pl — execution can lag the request by up to 24 hours. In KEV since 27 Sept. Fixed in 14.1-73.37 and 13.1-64.23. Patching does not remove existing implants. *CVSS v4.0 9.5 (Citrix) vs v3.1 9.8 (NVD). |
| CVE-2026-76443 | 9.8 | Cisco Secure Email Gateway and Secure Email and Web Manager, 15.5 and earlier | In the Wild (KEV, sibling) | No change in-window. Public Metasploit module (6.5.9, 6 Nov). Rapid7 describes the module as an S/MIME/email-parsing SQL injection leading to unauthenticated root RCE — treat the CVE mapping as Rapid7’s, not Cisco’s. The actively exploited zero-day in the same family is sibling CVE-2026-76461 (9.8, in KEV since 14 Sept). |
| CVE-2026-97229 | 10.0 | GitLab CE / EE | In the Wild (KEV) | No change in-window. Public Metasploit module (6.5.9, 6 Nov). CI/CD RegExp integer-overflow RCE as the git user; authenticated, requiring write access to a public project. In KEV since 29 Oct. |
| CVE-2026-119977 | 8.1 / Low* | WordPress core 4.7.0–7.1.3 | In the Wild (attempts) | No change in-window. 240,000+ blocked attempts (peaking at 74,000+ on 30 Oct). Still no victim publicly named and no confirmed compromise. Fix 7.1.4 / backport; block web access to wp-content/upgrade/ before updating. Public PoC: lysophavin18/CVE-2026-119977 (GitHub). A second repo claiming to bypass the 7.1.4 fix holds no code, only a download link — flagged as a likely malware lure, do not download. *Wordfence 8.1 vs WordPress Low. |
| CVE-2026-107441 | 8.8 (Zyxel) / 9.8* | Zyxel GS1900 series switches, ten models, web management interface | In the Wild (zero-day) | No change in-window. June fix added a check that ran after the unsafe copy in /cgi-bin/dispatcher.cgi, so June-patched firmware stayed exploitable. Payloads: Jitterly backdoor and Sixzut rootkit; suspected Chinese-speaking actor Kapibala/Red Heron (attribution via Binarly/Acronis, not independently verified). *Secondary analysis rates it 9.8. Not in KEV as of 7 Nov. Fix: 22 Oct firmware; change default credentials. |
| CVE-2026-109593 / 109594 | 9.1 / 6.8 | SonicWall Global Management System 9.5.2 and earlier (end-of-life) | Not reported exploited in-window | No change in-window. 109593 critical (9.1): unauthenticated information disclosure via internal file reads. 109594 medium (6.8): high-privilege SQL injection. SonicWall will not patch — migrate to Network Security Manager. |
| CVE-2026-72898 | 10.0 | Metabase 58+ (free and paid editions) | In the Wild | No change in-window. CSA Singapore warned 4 Nov that this SQL-injection flaw, fixed 6 Aug and on KEV since 11 Aug, is actively exploited and a PoC is public. Public PoC: 4minx/CVE-2026-72898 (GitHub), referenced in CSA alert AL-2026-158. Safe releases: 58.24, 59.21, 60.17, 61.11, 62.9, 63.5+. |
| CVE-2026-40217 | 8.8 | LiteLLM proxy | In the Wild (LLMjacking) | No change in-window. User-ID rebind to PROXY_ADMIN. Update LiteLLM, lock endpoints off the internet, rotate keys. |
| CVE-2026-98432 | 9.8 | Linux kernel IPv4 TCP, 4.4+ | Research exploit only | No change in-window. Demonstrated exploit needs MPTCP; RHEL fixed kernels require reboot, Ubuntu was still in -proposed on 3 Nov. |
| CVE-2026-58115 | 10.0 | Node-RED / Siemens IoT2050 | In the Wild | No change in-window. Port 1880 without adminAuth remains an active target. Siemens scope revision noted 8 Nov (more SIMATIC IoT2050 models). |
| CVE-2026-111497 | 9.2 | Google Brotli; Cisco and Splunk products (Node2Shell) | In the Wild (KEV) | No change in-window. Scope as of 5 Nov: 19 Cisco products affected, dozens under investigation, about a dozen Splunk products also affected. Same Brotli root cause also affects ClickHouse (CVE-2026-116366 below). KEV federal due date fell 10 November, but per the KEV rule a deadline alone is not a new development. Qualys published detection/verification guidance for PHP2Shell and Node2Shell on 12 Nov (four non-intrusive QIDs, two intrusive verification QIDs) and reports 167 million PHP installations across 10,000+ customers; attackers probed PHP2Shell four days before a patch existed. |
| CVE-2026-19868 | 7.5 / 7.1* | SonicWall SonicOS SSLVPN (Gen7 and Gen8 firewalls) | In the Wild (KEV) | No change in-window. Advisory SNWLID-2026-0014 updated to v1.1 on 9 Nov. A user holding valid credentials for an alternate authentication method can bypass SAML authentication controls on SSLVPN, and may also bypass SAML-enforced MFA. KEV 6 Nov, due 20 Nov. Fixed in 7.3.4-7013 / 8.2.3-8015. No workaround. *NVD 7.5 vs SonicWall 7.1. No verified public PoC — possible fake/malicious repos unverified, not linked. |
| CVE-2026-124079 | 9.8 | rConfig Core 8.0.0–8.2.18 / Pro before 8.4.1 | Not known exploited — no PoC confirmed | No change in-window. Deserialisation of untrusted data lets an unauthenticated network attacker execute arbitrary code via the Redis-backed Laravel Horizon queue. Fixed in Core 8.2.19 and Pro 8.4.1. |
| CVE-2026-58868 | Critical | Android 14–17 | Not known exploited | No change in-window (November Android security bulletin). Critical RCE without user interaction. No CVSS figure, exploitation evidence or PoC in sources read; bulletin date not independently verified from the primary bulletin. Flagged as thinly sourced. |
| CVE-2026-123551 | 10.0 | OpenAM 14.0.0–16.1.3 | Not known exploited — no PoC (vendor withheld) | No change in-window. OpenAM broadcasts live session tokens on the notification WebSocket when agent notifications are enabled (default); agent-only check bypassable unauthenticated. Fix 16.1.4. Workarounds published. |
| CVE-2026-123550 | 9.8* | OpenAM before 16.1.4 | Not known exploited — no PoC | No change in-window. SAML v2.0 SP accepts unsigned assertions on ECP/HTTP-Artifact bindings when WantAssertionsSigned is false. *cve.now lists 9.3; 9.8 is the figure used by most trackers. Fixed in 16.1.4. |
| CVE-2026-114432 | 9.8 | Contact Form 7 6.2–6.2.3 | Not known exploited — no PoC reported | No change in-window. Unauthenticated arbitrary file upload (RCE on stock Apache). 10,000,000+ installs. Fix 6.2.4. |
| CVE-2026-107331 | 9.9 | IBM DataStage on Cloud Pak for Data | Not known exploited — no PoC reported | No change in-window. Remote authenticated attacker can obtain sensitive information via absolute-path traversal. Sibling CVE-2026-107330 (7.7) arbitrary file write. Also in the 8 Nov IBM batch: IBM Concert CVE-2026-6715 / CVE-2026-15266 (both 9.8, unauthenticated OS command injection, fixed in 3.1.0). |
| CVE-2026-96652 / 96654 / 96655 / 96656 (Plex2Shell) | — | Plex Media Server on NAS devices | In the Wild (KEV) | No change in-window. Synology advisory SA-26:17 (8 Nov): fixed package 1.43.5.11079. Updating does not remove installed payloads. In KEV since 20 October. |
| CVE-2026-120984 | 9.1 | Red Hat Build of Keycloak, client management with authorisation services enabled | Not known exploited — no PoC | No change in-window. Authenticated attacker with high-privilege client-management access can upload a malicious JavaScript policy, leading to arbitrary code execution. No fix version listed in sources checked. |
| CVE-2026-118267 | 8.1 | WordPress WP Data Access plugin, 5.5.89 and earlier | Not known exploited — no PoC | No change in-window. Incorrect authorisation allows unauthenticated privilege escalation via key/val parameters on app/update. |
| CVE-2026-116366 | 9.5 | ClickHouse Server, via bundled Brotli 1.2.0 | Not known exploited — PoC unverified | No change in-window. Unauthenticated RCE through bundled Brotli 1.2.0 — same root cause as CVE-2026-111497. Fixed by Brotli 1.2.1. Aggregator-listed PoC/lab repo not verified — not linked. |
No new core-banking, SWIFT/payment-gateway or enterprise file-transfer CVE was reported in this window beyond the items above. CISA also published five Siemens ICS advisories on 12 Nov (industrial/OT, low BFSI relevance); one republishes Node2Shell CVE-2026-111497 for RUGGEDCOM APE1808 — a republication, not new.
From 10 Nov 12:45 UTC, an attacker using stolen GitHub credentials (PATs, bot tokens, Fly.io GitHub App tokens) opened probe pull requests — branch wfx-docs, titled “docs: dependency audit notes” — in 61,764 repositories via 4,837 compromised accounts. The payload steals GitHub Actions secrets, cloud credentials, SSH keys, shell history and AI service keys (OpenAI, Anthropic, Cursor, Copilot). At least 1,806–1,835 repositories ran the payload with credentials exfiltrated (SafeDep counts 8,093 commits pushed on public default branches). The attacker deletes runs within a minute, so gaps in run numbers are the tell. The payload leaked into Packagist digitalelvis/neuronai-studio v3.14.3 and Go modules (github.com/onuragtas/docker-env v1.0.278/280; gocurl; redock), plus npm pdf-html5-page-flip 2.0.1/2.0.2 (removed 12 Nov; GHSA-j3hx-5mxx-43qh). Exfil IP 193.32.204.199 is the same as the October GhostAction wave; accounts hit in October were significantly more likely to be hit again, and wfx-loot was injected into GhostAction workflow files in 100+ repos — but Aikido assesses the two waves are likely different actors. GitHub was not itself exploited; valid stolen tokens were used. The campaign was still active at last check, with 4,100+ new probes on 12 Nov alone (per OX Security via Corgea). Any bank or fintech with GitHub Actions secrets, cloud keys or AI keys in repos is exposed: rotate and audit run-number gaps. Confirmed by two independent analyses (Aikido + SafeDep).
See Threat Actor Activity for the full detail. The BFSI point: ScreenConnect is the remote-access tool MSPs use to manage client endpoints; many small banks and financial firms rely on MSPs. A compromised MSP is a direct path into financial-institution endpoints, and the session-cookie relay turns a technician’s own access into a spread vector — though Huntress stresses the technician-to-panel spread has not been observed to succeed. Huntress’s instruction: review your environment now even without an alert; look for unexpected mass activity in Run Command and Toolbox history and for an unauthorised second ScreenConnect client; do not connect to a suspected host through ScreenConnect (use EDR live response, WinRM or another out-of-band CLI); isolate the host and keep it isolated until cleanup is complete (isolation alone does not remove persistence, and systems have been observed reinfecting shortly after isolation was lifted); remove the rogue client, the TrapWatch task, C:\ProgramData\RelayTrap\ and toolkit files under %TEMP%, and the RelayTrapLiteAlert Run value; revoke exposed browser and identity-provider sessions. Block at firewall/DNS: 185.99.255.28, 155.94.150.39, newnewnew.gotdns.org, soxnewwork.homedns.org, sc.payproofkit.com, ns.mo7sen.dev, cdn007-digitaldownloads.com.
See Threat Actor Activity and the watchlist. The Legacy API condition is the key defensive discriminator: servers on Hotfix 2 with Legacy API disabled are not exposed, per N-able and Huntress. N-able has restricted some features/integrations, switched off the Intune integration as a precaution, and is not advising customers to take N-central offline. A follow-up hotfix is planned. American Banker noted on 11 Nov that this is N-central’s third security emergency since August, and many small banks’ IT providers (MSPs) use N-central to manage and access bank systems.
See Threat Actor Activity. The insurer/pension subsidiary of Caixa Seguridade confirmed registration data for all 7.6 million active customers was copied out. The company’s “no financial data” conclusion is preliminary, and the six-day gap between detection (6 Nov) and disclosure (12 Nov) is unexplained. Firms holding Caixa group data, and any firm with a similar group-subsidiary structure, should check which entity holds which customer data and whether a subsidiary breach reaches their customers.
An unauthenticated ClickHouse database run by the Latvian branch of SMS provider CSC Telecom sat open on the internet holding 125 million+ text messages sent Oct 2022–Sep 2026 (3m+ in Sep 2026 alone), updating in real time — so a watcher could in principle have read one-time codes and password-reset links before recipients. Contents included sender names, recipient phone numbers, full message text and routing metadata. Swedbank messages carried phone numbers, IBANs, amounts, payment recipients and account balances; SEB, insurers ERGO/Compensa/Artea, airBaltic, Latvia’s 112 emergency line and retailers were also senders. Cybernews found it 28 Sep, reported it 29 Sep; access was closed 27 Oct. Swedbank has stopped using the provider and opened an investigation; SEB says it used CSC only for general pension-fund information and sent no banking codes, payment details or passwords through it. CERT.LV requested an in-depth investigation and referred the case to law enforcement. CSC Telecom has not commented publicly (per Cybernews). 125 million is a message count, not a victim count. No evidence anyone other than researchers read the data; no confirmed account takeovers or financial loss. The exposure was at a third-party SMS aggregator, not in either bank’s systems — a supply-chain / OTP-channel risk story, not a bank breach. Banks should ask their SMS aggregators how message stores are secured and whether OTPs are logged.
Hyakujushi Bank (Takamatsu) said 1,500 shareholder email addresses leaked after unauthorised access to the mail-delivery system of Arara, used by a contractor that sends e-gift prizes to shareholders. The addresses were those sent e-gift winning emails in July 2026. No shareholder numbers, other attribute data, or email body content were included, per the bank. No misuse confirmed so far; the bank warned about impersonation emails. This is fallout from the Arara breach, not a new intrusion at the bank. Arara’s 6 Nov third report put the total at 81,329,467 email addresses (deduplicated) across up to ~380 client companies, exfiltrated 28–31 Oct; the same platform was also abused 28–30 Oct to send up to ~7.13m phishing emails impersonating a credit-card company. Other 11–12 Nov Arara fallout: MUFG (~20,000 shareholder emails) and Nomura (~13,500) disclosed 11 Nov; Japan’s Cabinet Office ~62,863 newsletter registrant emails; Okayama Prefecture 2,241 emails disclosed 12 Nov. Do not add these figures together — they are subsets of the same Arara incident.
Recorded Future disclosed that Klue, a third-party marketing vendor, had unauthorised access to the integration layer linking it to Salesforce and other marketing/sales SaaS. Activity began 12 June 2026 and was contained the same morning; on 17 June Recorded Future confirmed elements of its Salesforce account were affected via a compromised OAuth token. Impact limited to business data in Salesforce (client contact names/emails, some contract info); no evidence its core platform, Intelligence Graph or internal infrastructure was touched. Tokens revoked, Salesforce engaged, law enforcement notified. Timing caveat: the underlying incident is June 2026 and the blog page carries no publication date — only the disclosure surfaced in-window. The pattern (compromised OAuth token into a Salesforce integration) echoes the AU Small Finance Bank item below and is the lesson, not a new incident.
See Threat Actor Activity. The bank is investigating a possible breach of limited customer data linked to its vehicle-loan portfolio, hosted on a third-party platform operated by Salesforce. Core banking systems not indicated as compromised. Reported / unconfirmed — the bank has not detailed the data involved.
Singapore’s Personal Data Protection Commission and CSA published a joint advisory on data-protection risks arising from insecure APIs. Guidance, not an incident, and no CVE. BFSI-relevant because APIs are core to banking and fintech (open banking, payment APIs): the advisory is a prompt to review API authentication, authorisation and data minimisation, in the same week the SinoPac incident showed what a vendor middleware weakness can do.
Notion disabled exposed API keys on 11 Nov after finding them inside Notion AI chat transcripts exposed via the Fireworks AI breach; it had disconnected from Fireworks on 3 Nov, deleted Fireworks-held data, disabled its Insights feature and stopped serving open-weight models through Fireworks. Fireworks notified Notion of potential customer-data access on 6 Nov and confirmed the dataset (including Insights transcripts) on 8 Nov. Fewer than 1% of workspaces affected; no misuse of the revoked Notion keys reported. Fireworks attributes the October intrusions to a financially motivated LLMjacking crew reselling stolen model access; billing data, secrets, datasets and incomplete model weights were exposed for a limited set of accounts. This is a follow-up development on the Fireworks/Notion story already covered on 11–12 Nov, not a new breach. The standing lesson holds: treat anything pasted into an AI chat as potentially exfiltrated, and rotate keys.
FinCEN’s order takes effect 10 December 2026. Binance is already rejecting MEXC-bound withdrawals and holding MEXC-origin deposits for compliance review. The “at least $16 billion” (Treasury) vs “$16.7 billion” (FinCEN order text) figure conflict is unchanged.
Wiz/Mandiant disclosure was 9 Nov; AWS’s statement (“AWS services were not compromised”) was added 10 Nov (pre-window). IOCs from that disclosure remain: Qubole AWS account 805246085872; IAM principals AIDA3W7DIFLYOLTK63SFS and AIDAJYRRH5Y3A7AAJM2HW; quiam- role-session names; IPs 57.129.54.120 (OVH), 45.83.220.211, 185.209.199.73, 89.37.63.175 (Mullvad exits); sso_agent.iso SHA-256 65f061281c9a712e9502793dc4c5035cae490b84da05f7d4223cefa65ad396e4.
Thai Airways said 9 Nov that ~60,529 passenger records were accessed via Amadeus (Aug 2023–Sept 2026); a Finnair figure of ~116,000 bookings surfaced only via a secondary roundup. No new Amadeus statement in-window. BFSI relevance indirect (travel PII usable for phishing; no payment data reported taken).
Client letters mailed 9 Nov; Texas AG portal lists 10,456 affected Texas residents, with SSNs, full names, addresses and other information. The third-party provider remains unnamed.
See Threat Actor Activity. If true, a supply-chain risk to US credit unions and community banks via their digital-banking provider; treat as a watch item, not a confirmed breach.
Europol and law enforcement from nine countries (Bulgaria, Canada, Denmark, Germany, Moldova, Netherlands, Poland, UK, US), coordinated with Europol/Eurojust and private partners, seized 235 servers and 257 domains behind Gootloader, CastleRAT and StealC. Breakdown: Gootloader 14 servers / 218 domains (681 infected websites cleaned), CastleRAT 120 servers / 39 domains, StealC 102 servers. Authorities secured 6.4 million victim datasets / credentials harvested for sale and reuse in further attacks including ransomware. CastleRAT victim data will be shared with Have I Been Pwned, DIVD, CheckjeHack, NoMoreLeaks, Sekoia, Shadowserver and NL-NCSC for victim notification. No arrests were reported in the coverage seen. Small discrepancy — flag: the per-family server figures sum to 236 vs the stated 235 total. StealC is an infostealer harvesting banking credentials, saved logins and crypto wallets; Gootloader is a ransomware initial-access vector; the seized credential trove feeds account takeover and fraud. Check staff and customer credentials against Have I Been Pwned notifications; treat StealC-sourced session tokens as live account-takeover risk.
A 45-year-old woman from Kozhikode lost ₹49.2 lakh between 15 July and 22 September after being lured by a fake Facebook video purportedly showing Union Finance Minister Nirmala Sitharaman promoting high-return online trading. She transferred funds from State Bank of India and Puthuppadi Service Cooperative Bank accounts to accounts supplied by the fraudsters. Kozhikode Rural Cyber Crime Police registered a case on Wednesday 11 Nov and are tracing/freezing the recipient accounts; police warned against AI-generated videos of ministers promoting investment schemes. This is the clearest in-window AI-scaled fraud item — a confirmed police case with an AI-deepfake element as reported by police/PTI, not a new campaign. No new AI-scaled phishing, vishing or deepfake campaign was found in-window beyond this case.
Assam Police arrested two men in Guwahati for operating a cyber-fraud network using mule bank accounts and linked merchant/personal QR codes to collect, layer and cash out scam proceeds from victims across India. One arrestee, Subham Sah (27), is a former Temporary Business Development Executive at IDBI Bank who allegedly abused insider knowledge of banking procedures to open multiple mule accounts in other people’s names; the other, Shamnur Rahman (33), was intercepted at an IDBI Bank branch in Panbazar while trying to reactivate a dormant account. Police are examining KYC documents, devices and transaction logs to trace beneficiaries; scale and total amount not disclosed. Direct BFSI relevance: insider-facilitated mule onboarding and QR-code cash-out rails — dormant-account reactivation and staff/ex-staff knowledge are mule-network enablers.
A fraudster posing as a Karnataka senior police officer and CBI officer told a 71-year-old Pune man his SIM had been misused for child trafficking, threatened “digital arrest”, and persuaded him to transfer ₹22 lakh as a “security deposit” to accounts provided. The fraud ran 5–15 October; the victim complained on Wednesday 11 Nov and Vishrantwadi police registered a case under BNS cheating-by-personation provisions and the IT Act. No arrest yet. Classic government-official impersonation / digital-arrest script ending in transfers to mule accounts; same typology as the SPF/MAS warning already covered. The fraud itself predates the window — the reporting is in-window.
Bermuda Police warned of callers spoofing Bermuda (+1 441) numbers, posing as a local bank, falsely claiming fraudulent activity on accounts and requesting login credentials, then rapidly transferring funds overseas. Three numbers named: +1 (441) 735-2712, +1 (441) 737-5993, +1 (441) 905-4600. Police reiterated banks never ask for login credentials by phone. Small scale, direct relevance — bank-impersonation vishing with credential theft and fast overseas cash-out.
Ahmedabad Cyber Crime Branch warned of social-media messages falsely claiming PM Narendra Modi is offering ₹8,000 Diwali cash aid. Links lead to fake sites that harvest a WhatsApp “Connection Code” (Linked Devices) to hijack WhatsApp/Telegram accounts, and push malicious APKs purporting to deliver the reward via PhonePe, which can access SMS, OTPs and banking apps to drain accounts. Victims are told to forward the message to WhatsApp groups to spread it. A current, actively spreading lure: smishing/social-engineering chain ending in OTP and banking-app compromise.
See Threat Actor Activity. Over 50 traders and customers in West Kordofan claim millions of Sudanese pounds were stolen from Bank of Khartoum accounts via the Bankak mobile banking app, with individual reported losses of SDG 18–45 million; one customer reported SDG 21 million taken in two transfers (14 Oct and 3 Nov), another SDG 12 million on 14 Oct without opening links or messages. The Dar Hamar Emergency Rooms documented the cases, demanded reimbursement/investigation from the bank, the Central Bank of Sudan and prosecutors, and urged a boycott. Bank of Khartoum had not responded to comment requests at publication. No evidence presented that the app itself was breached vs credential theft or social engineering; the theft dates predate the 12 Nov reporting. Treat as an allegation, not a confirmed bank breach. Context: Bankak is Sudan’s dominant payment channel amid the war/cash crisis, so trust impact is high even if unproven.
The New Indian Express (12 Nov) reports that CERT-In and the Ministry of Home Affairs advised banks, NBFCs and payment aggregators “earlier this month” about AI agents micro-testing compromised bank accounts (₹1 test transactions) before draining them. The same piece describes digital-arrest video calls and AI voice cloning as surging scam types in India, in general terms. The advisory itself predates the window and no exact date is given in the piece, so treat as reported context, not a new in-window advisory. The ₹1 test-transaction pattern is a concrete detection signal for fraud teams either way.
Twelve new unauthorised-entity warnings: FCA 8, ASIC 3, FINMA 1. Notable: a clone impersonating licensed firm Eley Griffiths Group (uameshx.com, ASIC warning 11 Nov); FINMA warning on bsd-capital.ai (10 Nov); FCA warnings including marketexpertgroup.co.uk and divinegold.co.uk (6 Nov) offering CFDs/forex/digital assets without authorisation. Confirmed regulator warnings via a secondary roundup (underlying warnings dated 6–11 Nov). Routine list hygiene rather than a new campaign — low to medium relevance.
Fraudsters send messenger messages impersonating the Federal Tax Service, telling recipients to verify tax charges via a link to a fake site, then claim the victim’s Gosuslugi (state services) account was hacked. The ministry also warned (11 Nov) that scammers posing as officials/bank staff/lawyers target people who have already lost money, offering fake fund-recovery help. Confirmed ministry warning, translated from Russian — wording nuance flagged. Same recovery-scam pattern as earlier HKMA warnings. Low to medium relevance.
Caixa’s leaked CPFs, names and contact details are impersonation material for 7.6 million customers — expect phishing citing Caixa Vida e Previdência or Caixa Seguridade. Arara’s 81 million leaked shareholder emails are live phishing bait, and the same platform was already abused to send ~7.13m phishing emails impersonating a credit-card company. Seven Bank’s residence-card data is identity-document fraud material. FSI leak gives attackers real assessment-staff contact details and firm names — expect phishing citing the FSI or the evaluation process. SMBC Nikko’s leaked holdings and account details are impersonation material for shareholders. SMBC CF’s SMS from 0005000365 will be spoofed. Philippine SEC notices remain bait; the SEC warns it never asks for passwords or security PINs. Mirae Asset’s leak gives attackers 3,800 customer account numbers plus names, emails and mobiles of 35,255 employees. Binance, Mercado Bitcoin, Gopax and Nubank/Mercado Pago outage messages remain live phishing bait. Verify via official status pages only.
No DDoS-extortion reporting against banks or financial firms dated in-window was found in the searches run. Absence of reporting is not evidence of absence of activity.
Techniques actually published for this edition. No mapping is invented where a source published none, and older research recirculated in-window is marked as context, not as a new sighting.
| MITRE Technique ID | Technique / behaviour | Seen in this edition | BFSI relevance / detection note | Status |
|---|---|---|---|---|
| No mapping published | Third-party RMM extension accepts a blank access key, bypassing ScreenConnect accounts, roles and MFA; compromised panel’s “Run Command” pushes a rogue client + SCKit toolkit; RelayTrap steals the technician’s session cookie (SQLite read of browser cookie databases) and replays it against the panel; Shared Toolbox upload reaches every session the technician can access; scheduled task “TrapWatch” (SYSTEM) | ScreenConnect SCKit worm — Huntress, 12 Nov | Check RMM+ versions 1.0.31/1.0.32 and re-key or remove the extension; hunt for RelayTrap, TrapWatch, the rogue client IDs and the new IOCs; do not connect to suspected hosts through ScreenConnect. The technician-trap spread was designed but has not been observed to succeed. | Confirmed (vendor research) |
| No mapping published | Apache Axis CVE-2026-90436 → setLDFileOwner symlink privilege escalation (VULN-2026-0009) → root on the N-central server; XMRig miner pulled from 157.20.241.16; probes to /dms/services/ServerUI referencing jline.internal.TerminalLineSettings; files under /tmp/.xvd; Agent2Shell chains 10+ flaws to SYSTEM on managed Windows agents | 4EverAble2Shell / Agent2Shell — 12 Nov | Disable the Legacy API (non-default) — servers on HF2 with it disabled are not exposed. Hunt for miners, xvd.service, /tmp/.xvd and ServerUI probes; treat a hit server as fully compromised. | Confirmed disclosure; exploitation claimed, contested (N-able vs Previdian/Huntback) |
| No mapping published | Autonomous AI intrusion: target domain + one-line goal; shared-blackboard fact/intent loop plans and executes; known weaknesses only — auth bypasses, IDOR, mass-assignment privilege escalation, weak/default credentials, exposed cloud APIs, dated CVEs; persistent admin accounts and API tokens created in production systems; unauthenticated cloud-storage writes | Cairn (CL-STA-1238) — Unit 42, 12 Nov | Identity hygiene, attack-surface reduction and patching still work — autonomy raised volume and coverage, not complexity. Hunt for new admin accounts and API tokens, and unauthenticated write access to cloud storage. | Confirmed (vendor research); attribution moderate confidence |
| No mapping published | PHP2Shell in the Mobile Access Portal + Gaia DynamicCLI local privilege escalation (CVE-2026-123039) → root on the gateway; PHP webshells in login templates (zz*.tpl, key ‘zz2026’); persistent TSGA backdoor with TSProxy component; gogo scanner for internal discovery; SSH pivot to a VMware ESXi host; destroy_v6 wiper; DREAM_ANNEXATION.txt notes | Check Point PHP2Shell gateway exploitation — 12 Nov | Apply the November Jumbo Hotfix; hunt for the TSGA backdoor, upload_progress_pwn in PHP session files, CPconfigdaemon masquerade processes and wiper/notes files. The PHP2Shell fix had been available since 23 Oct. | Confirmed (vendor research); attribution not high-confidence |
| No mapping published | Stolen GitHub credentials (PATs, bot tokens, GitHub App tokens) open probe PRs (branch wfx-docs, “docs: dependency audit notes”) that steal GitHub Actions secrets, cloud credentials, SSH keys, shell history and AI service keys; runs deleted within a minute, leaving gaps in run numbers; payload leaked into Packagist, Go modules and npm packages | wfx-loot — Aikido, 12 Nov | Rotate Actions secrets, cloud keys and AI keys; audit for run-number gaps and wfx-docs PRs; check pinned package versions against the affected releases. | Confirmed (two independent analyses) |
| No mapping published | Video-call lure on imposter Zoom / Teams / Meet; webcam video secretly recorded, then used to target the victim’s contacts; AI deepfakes of well-known crypto figures now being used | FaceStealer (Sapphire Sleet / UNC1069) — FBI, 12 Nov | Verify video-call identities out of band before acting on payment or access requests; brief staff that recorded video can be reused against their contacts. | Confirmed (government announcement) |
| No mapping published | Fake meetings, shared financial documents, fake Chrome PDF-viewer updates, fake media players, identity-verification prompts, hardware-wallet access and security-audit workflows; AI deepfakes (Google Generative AI, Grok Imagine, Wondershare AIGC, GPT-4o) no longer distinguishable by visual inspection; intrusion still depends on the victim running a “fix” | BlueNoroff — Arctic Wolf, 12 Nov | Verify meeting and document lures through a second channel; treat any “fix” or update prompt arriving with a meeting invite as hostile until proven otherwise. | Confirmed (vendor research) |
| No mapping published | Memory flaws in NetScaler ADC/Gateway (up to 9.5); unauthenticated Console authentication bypass (9.4 / 9.3 — see watchlist conflict); Secure Access Client privilege escalation to SYSTEM | Citrix NetScaler November batch — 12 Nov | Patch ADC/Gateway to 14.1-73.57 / 13.1-64.39 and Console to 14.1-74.34; Console bypass is the priority (management plane). | Confirmed (vendor); no exploitation or PoC reported |
| No mapping published | Race condition in mod_http2 assign_headers() corrupts shared heap data → remote code execution; public PoC now exists | Apache CVE-2026-119618 — PoC published 12 Nov | Upgrade to httpd 2.4.71; disabling HTTP/2 is a mitigation. PoC availability raises urgency; in-the-wild exploitation not confirmed. | Confirmed (vendor); PoC verified genuine research code |
| No mapping published | Origin-validation flaw → unauthenticated remote code execution on the endpoint agent | FortiClient CVE-2026-121529 — catch-up (CVE 10 Nov, exploit 11 Nov) | Patch FortiClientMac / FortiClientWindows to fixed releases; public exploit exists. | Confirmed (vendor + public exploit); in-the-wild exploitation not reported |
No technique in this edition was published with a MITRE ID in sources read. (BlueNoroff’s MITRE mapping sits in the full report PDF, which was not fetched — no IDs are invented here.) Recurring-TTP watch note (context only, not a new sighting): RMM and MSP compromise is the thread of this edition again — ScreenConnect and N-able both turn the tool that manages bank endpoints into the attack path, now in a second N-able wave three days after the first — and credential theft at scale recurs across it (wfx-loot’s CI and cloud keys, SCKit’s session-cookie relay, Cairn’s API-token creation, the FSI’s fake regulator account). Hunt for unusual service-principal changes, first use of cloud keys from new IPs, gaps in CI run numbers, and RMM sessions outside change windows. The FSI official’s “AI use likely” estimate is not confirmed AI use, and the Bank Rakyat AI speculation remains an unattributed opinion post — neither is counted as a confirmed AI technique here.
AI as a target, AI as a weapon or accelerator, and AI as a risk and control problem for BFSI.
Notion revokes API keys after the Fireworks breach (follow-up, 12 Nov). Notion disabled exposed API keys on 11 Nov after finding them inside Notion AI chat transcripts exposed via the Fireworks AI breach. Fireworks attributes the October intrusions to a financially motivated LLMjacking crew reselling stolen model access; billing data, secrets, datasets and incomplete model weights were exposed for a limited set of accounts. wfx-loot (NEW, Aikido, 12 Nov): the GitHub CI secret stealer specifically harvests AI service keys (OpenAI, Anthropic, Cursor, Copilot) alongside cloud credentials and SSH keys — at least 1,806 repositories ran the payload with credentials exfiltrated. Recorded Future Insikt Group AI report (coverage 12 Nov): 103 AI-linked incidents, 31 Jul 2025–31 Jul 2026 — 32 involved AI in malware/offensive operations; 71 were attacks on AI infrastructure, AI-themed delivery, AI-assisted deception or abuse of AI services. Key cases: the LiteLLM PyPI compromise (Feb 2026, malicious v1.82.7/1.82.8), the Shai-Hulud worm expanding to Claude Code (May 2026), and a state-sponsored campaign against Taiwanese government targets using up to eight AI agents (Hermes/OpenClaw framework; 85 credentials cracked, 2,500+ personnel records exfiltrated, July 2026). Attackers mostly use AI to accelerate familiar crime (supply-chain compromise, phishing, extortion); fully autonomous operations remain rare. Timing caveat: the report appears to have gone live ~12 Nov 00:35 SGT, just before the 09:00 window — the in-window item is the 12 Nov coverage. Standing: Google/Mandiant LLMjacking research (2–3 Nov) — stolen cloud credentials and LLM API keys, fraudulent bills over $100,000 in a single day and up to ~$1m cumulative at some victims, financial services among victims; Railway’s final incident report (10 Nov) — ~$740,000 in unauthorised AI usage on stolen keys. Standing controls: key inventory and rotation, endpoint exposure, spend and invocation monitoring with revoke-and-rotate.
Cairn autonomous intrusions (NEW, Unit 42, 12 Nov). A Chinese-speaking operator ran 25 largely autonomous operations against at least 19 organisations in Aug–Sep 2026 using the open-source Cairn framework, supplying only a target domain and a one-line goal; the framework planned and executed itself, paired with Zhipu GLM 5.2 and DeepSeek models. Unauthorised access or data retrieval was confirmed at 10 organisations. Unit 42’s point stands: autonomy raised volume and coverage, not complexity. BlueNoroff (NEW, Arctic Wolf, 12 Nov): AI deepfakes built with Google Generative AI, Grok Imagine, Wondershare AIGC and GPT-4o, no longer distinguishable by visual inspection, used to lure crypto and venture-capital decision-makers. FaceStealer (NEW, FBI, 12 Nov): North Korean actors are starting to use AI to create deepfake videos of well-known crypto/blockchain figures as part of a scheme that has stolen over $400m. Kozhikode deepfake case (NEW, police case reported 12 Nov): a fake video of India’s finance minister lured a victim into a trading scam, ₹49.2 lakh lost — the only in-window AI-scaled fraud case found. No new ARTEX, CyberStrikeAI, AntiFox, PentAGI or HexStrike AI reporting, tool update or observed use dated 12–13 Nov was found (the latest material is 5–11 Nov); Cairn is the new agentic item, and it is a different framework. Check Point’s October 2026 threat landscape (11 Nov) — ransomware hit a record 1,063 reported attacks in October, and new extortion group “n0n” entered the top 3 within a month, “relying heavily on AI-assisted tooling” — vendor telemetry, and the AI-tooling claim is Check Point’s characterisation, not independently verified. No new AI-scaled phishing, vishing or deepfake campaign was disclosed in-window. The $129m AI voice-clone indictment is a 5 Nov unsealing — fraud-pattern warning, not new.
Post-training supply-chain attacks on coding agents (NEW, Microsoft, 12 Nov). Microsoft Security Blog (Defender Research + AI Red Team), “Trained to betray”: two studies show hidden backdoors in coding/terminal agents via the post-training supply chain. Dataset poisoning — 90 crafted examples hidden in ~9,000 fine-tuning conversations (~1%) taught a gpt-4.1-mini model to find any available command-execution tool and misuse it on routine, even paraphrased, requests (~78% attack success on in-distribution harnesses, transferring to an unseen harness). Model poisoning — an attacker controlling training before distributing an open-weight checkpoint (Qwen3.5-4B/9B, SFT + RL) reached 84.1–94.8% success on unseen repositories while preserving benign performance, with backdoors surviving later training. A dormant backdoor in an agent can invoke tools, execute commands, modify files or access data when triggered. Recommended controls: training-data scrutiny, checkpoint provenance, deployment-specific evaluation, restricted tool permissions, sandboxed execution, runtime monitoring. This is research, not observed attacks — report it as such. FCA AI-review analysis for payments firms (NEW, 13 Nov 07:59 SGT). Analysis of the FCA’s Wholesale Banks Supervision Insights 3 (2 Nov) and the joint Bank of England/FCA AI survey (196 firms; 86% of FS firms now use AI, 94% have a firm-wide AI policy, human review the most common control at 95%). It recaps FCA findings already reported: AI-generated code errors passing human review and causing a loss of service at one wholesale bank, staff use of unapproved AI tools leaking sensitive data, and untested kill switches. Recommendations: tested kill switches with named authorised operators, accurate AI inventories, disclosure amnesties for unapproved AI use, and treating unapproved AI as a standing board item. The FCA will not write new AI rules but expects senior-manager accountability and incident reporting to extend to “significant AI failures and near-misses”. The 12 Nov date is the analysis, not new regulator action. Standing: OCC Bulletin 2026-54 (released 4 Nov; analysis 11 Nov) creates no new rules and sets no supervisory expectations; Digidentity survey (11 Nov): 69% of UK banking customers doubt their bank can detect AI deepfake impersonation; 80% expect banks to use advanced deepfake detection; 53% say SMS one-time codes are no longer sufficient (survey data, not an incident). AI-assisted vulnerability discovery is now a standing fact on both sides: Fortinet’s two new FortiMail criticals were found with the Frontier AI model Claude Mythos; Microsoft’s FORGE Lab (11 Nov) reports AI-assisted scanning produced 33 kernel CVEs; Tenable notes the exploited Outlook CVE-2026-124194 was discovered by a researcher using Claude; Citrix’s 12 Nov NetScaler fixes followed a deeper codebase review using AI-assisted code analysis. Standing: NCSC UK chief executive’s 9 Nov warning of an AI-powered ransomware attack on UK critical national infrastructure being “highly likely” by 2028 is a forecast/assessment, not a new incident. Standing controls: CSA/GovTech/IMDA personal-AI-agent advisory (6 Nov) — do not give agents access to banking, government or healthcare accounts; Sophos malicious AI-agent “skills” research (6 Nov); Haiou “prompt adware” (Check Point, 4 Nov); Elastic’s finding that attackers execute commands through Claude Code and ChatGPT desktop in confirmed intrusions. Also standing: the npm REF8801 packages that rerouted AI-agent traffic through the operator’s API (@apexacc/cli replaced with a security placeholder 11 Nov; @apexfdn/copilot-mcp still published — treat as attacker infrastructure).