◉ BFSI CTI Briefing — Open-source synthesisWindow: 31 Oct – 1 Nov 2026 • Asia / Singapore
Critical — Act todaySunday 1 Nov 2026 • SGT evening24-hour window

The laundering trail, the vendor script, and the 1 Nov patch deadline.

This briefing covers reporting dated 31 October – 1 November 2026. A few items were first disclosed on 29–30 October. They appear only where a deadline, update, attribution or follow-up fell inside the window, and they are labelled as such. Each finding below states what is confirmed, what is disputed, and what is still unknown. No new FIN7, Scattered Spider, Carbanak, Akira, RansomHub or Cl0p attack on a BFSI victim, no FI-specific IAB sale, and no DDoS-extortion attack on a bank was published in the window. That means none was found in the sources searched. It does not mean none happened.

~$6.3–6.4bnBinance theft attributed to TraderTraitor — sources vary $6bn / $6.3bn / $6.4bn
628FBI wallet addresses to block — 608 ETH · 19 BTC · 1 SOL
10.0 × 2Two GitLab CVSS 10.0 RCEs — only one is in KEV
1 NovKEV due date reached — Cisco ASA/FTD & GitLab CVE-2026-97229
01 — Executive Summary

Three things that matter most

A short summary only. The detail, and the limits of each claim, are in the sections below. Do not act on a headline figure without reading its limits.

1

State-sponsored crypto theft is the biggest event — act on the laundering trail now

On 31 Oct the FBI attributed the ~$6–6.4bn Binance theft of ~29 Oct to North Korea’s TraderTraitor (Lazarus cluster) and published 628 wallet addresses for blocking. The actors are converting proceeds mainly to ETH and moving them through DeFi and instant swaps. Binance said it had replenished ~90% of losses, frozen ~$3bn with partners, and planned to reopen withdrawals on 1 Nov.

Critique: the attribution is a formal law-enforcement claim, but the evidence behind it is not public in full, so it cannot be checked independently here. The loss total is not settled — sources give $6bn, $6.3bn and $6.4bn. The replenishment and freeze figures come from Binance and its partners, not from an audit. And the FBI release does not explain how the DevOps laptop was compromised, so the way in is still unknown.

2

The data incidents came through vendors, not through core banking systems

On 31 Oct, NTT Data suspended STdirectServices after unauthorised access to an unnamed subcontractor’s SaaS platform exposed Japanese bond buyers’ names and purchase amounts (Rakuten up to 362, SBI Securities 1,169, MUFG Morgan Stanley 95, Kintetsu total reported 2,678 — these counts overlap and must not be added together). Separately, AU Small Finance Bank said it was investigating a possible leak of limited vehicle-loan data on a Salesforce-hosted SaaS platform, after an unverified ScarlettGroup claim.

Critique: in both cases the firms say core systems were not accessed. That is their own statement. It has not been independently verified, and the subcontractor in the NTT case is not even named. The AU claim may not be a breach at all. What is clear is the fraud risk: phishing and vishing that quote real names, amounts and loan details. That is a risk of misuse of real data, not proof of account takeover.

3

The 1 November KEV deadline is the patching emergency — and the Ivanti chain is the trap

Cisco ASA/FTD CVE-2026-76418 (8.6) and GitLab CVE-2026-97229 (10.0) both had a 1 Nov federal remediation date. Ivanti Connect Secure CVE-2026-111357 is due 2 Nov and CVE-2026-118527 on 13 Nov. The Ivanti flaws chain with the 13 Oct CVE-2026-104935 to reach root, and the chain was used against a fully patched appliance.

Critique: a passed deadline does not mean every system is now safe. It only means the date has passed. A GitLab upgrade alone does not fix custom-Ruby builds, and a passing Ivanti Integrity Checker result does not prove an appliance was never touched. Patching without a compromise hunt is not enough.

⚠
Deadline tracker — read before the SOC stand-up

31 Oct due: Linux kernel CVE-2026-46300  ·  1 Nov due: Cisco CVE-2026-76418, GitLab CVE-2026-97229  ·  2 Nov due: Ivanti CVE-2026-111357  ·  11 Nov due: VMware ESX CVE-2026-41709  ·  13 Nov due: Ivanti CVE-2026-118527  ·  PAN-OS PHP flaws were due 29 Oct.

02 — Threat Actor & Campaign Activity

Who did what — and what is only claimed

If a source published no MITRE ATT&CK list, the card says so. No mapping has been added. Check the status tag on each card before acting — a claim and a confirmed incident need different responses.

TraderTraitor — North Korea, Lazarus clusterFBI attribution · 31 Oct 2026

Confirmed attribution

Crypto exchange / virtual-asset finance, global — Binance.

No MITRE technique list was published in the 31 Oct FBI release summarised in the research.

As reported: compromised DevOps engineer’s laptop → service container → unauthorised withdrawal submissions; abuse of Ethereum EIP-7702 pre-signed authorisations (~50, per TRM / secondary reporting) to sweep hot wallets across 10 chains in ~20 minutes; conversion mainly to ETH, then DeFi / instant-swap laundering incl. THORChain.

Context outside window: fake-job-interview / “Contagious Interview” social engineering reported 27 Oct.

Largest crypto theft on record: ~$6.3–6.4bn claimed (sources vary; TRM ~$6.36bn). ~$2.7–3bn frozen by issuers/partners; ~$2.4bn still attacker-controlled per TRM.

Binance said it would cover user losses from reserves and reopen withdrawals 1 Nov. No private-key leak found, per Binance.

Critique: the “no private-key leak” finding and the recovery figures are Binance’s own statements. They have not been audited, and reserves claimed are not the same as liabilities proven.

ScarlettGroupReported 30–31 Oct 2026

Unverified claim

Banking, India — AU Small Finance Bank, vehicle-loan customers, via Salesforce-hosted SaaS.

No MITRE mapping published in sources.

Claimed access to KYC, PAN/Aadhaar, credit reports, bank statements, loan agreements, CRM, employee and audit-log data, with threat to leak.

Unconfirmed breach. The bank confirms only that it is investigating. Salesforce (31 Oct) says no Salesforce platform vulnerability was identified. The bank states core systems are unaffected and services are normal. The number affected and the data volume are undisclosed.

Critique: a leak claim is not proof of a breach. Treat this as an extortion or credibility claim until forensic evidence is published. Salesforce’s denial also has a limit: no platform flaw found does not rule out stolen credentials or misuse of a customer account. The fraud risk is real only if the KYC data is genuine, which is not yet known.

Unknown actor via DebugBear — Polymarket supply chainIncident 29 Oct · SecurityWeek update 31 Oct 2026

Supply-chain compromise

Prediction-market / trading finance, international platform (US platform reportedly unaffected).

No MITRE list published in the Hudson Rock excerpt.

Malicious JavaScript served from a compromised third-party RUM script; hunted logged-in users’ trading tools / portfolio in the browser; sold positions and moved proceeds, or sent position tokens to an attacker-set address; reused already-authenticated session authority. Hudson Rock found no routine sending passwords, seed phrases or private keys.

~$2.19–2.71m drained (estimates vary by researcher; Polymarket disclosed no total). ~700 wallets / 710 deposit wallets and 731 addresses in on-chain analysis (address count ≠ victim count). Polymarket halted international trading ~1 hour, pledged full reimbursement, referred to law enforcement. Second such vendor-script compromise for Polymarket in 2026 (June ~$3m).

Critique: every loss figure here is a researcher estimate. Polymarket has published no total of its own and has not named the vendor, so the true loss and the full victim count are not known. A reimbursement pledge is also not the same as reimbursement paid.

Unknown intruder — NTT Data STdirectServices subcontractorDisclosed 31 Oct 2026 · Incident dated 30 Oct by Rakuten

Third-party incident

Securities / brokerage, Japan — bond investors via Rakuten, SBI Securities, MUFG Morgan Stanley.

No MITRE mapping, actor, or method published. Unauthorised access to a subcontractor’s SaaS “digital securities platform” used for bond-perk applications.

Names + purchase amounts for bond buyers; for 2 JR West investors also address, phone, email, gender, DOB. Login IDs, passwords, account numbers and trading details stated as NOT included; no unauthorised logins/trades found, per Rakuten. Primary risk: highly convincing phishing / vishing quoting a real bond and amount.

Critique: the statement about what was not taken comes from the firms involved. It has not been independently checked. The subcontractor is unnamed and the method and start date are unknown, so the full exposure cannot be judged from this disclosure alone.

Vishing / impersonation actors — Butterfield & Bermuda warningButterfield updated 31 Oct · Bermuda Monetary Authority 30 Oct

Fraud — vishing

Banking customers, Cayman Islands / Bermuda.

No MITRE list published. Phone impersonation of bank/regulator; claim of attempted hack or frozen account; second call from fake “cybersecurity expert”; direction to fraudulent live “support chat” link; request for device/computer access and personal/banking details.

Victim accounts later raided (Butterfield cases); contacts bombarded with messages appearing to come from the victim. Banks/regulators state they never ask customers to click a link, transfer money, install software, or give banking details by phone.

Critique: these are victim and bank reports of a fraud pattern. No total loss figure has been published, so the scale of this campaign is unknown. The pattern is still worth warning customers about, because each reported case follows the same steps.

Payment-diversion fraud — Capillary / SessionM Czech RepublicKPMG findings updated 31 Oct 2026

Fraud — €3.04m

SaaS / loyalty-platform subsidiary; banking-payment fraud, €3.04m.

No MITRE mapping published. Social-engineering-led: impersonation of senior executive and external lawyer, lookalike email domain, WhatsApp and personal-email communications. KPMG found NO evidence in examined records of malware, unauthorised access, mailbox compromise, account takeover or traditional email spoofing.

€3.04m in 4 disputed transactions; €450k recovered. 3 payments to ZZDJB CO LIMITED at HSBC Hong Kong; 1 to ZHONGXIAN TRADING PTE. LTD at UOB Singapore. Ultimate beneficiaries and onward movement not traced in findings reported.

Critique: KPMG found no technical compromise in the records it examined. That is a limit of the records checked, not proof that no system was touched anywhere. The named beneficiaries are where the money was sent, not proof of who ultimately received or kept it.

LockBit — ransomware cartel (RaaS)MPCERT advisory · 1 Nov 2026

Advisory — not a BFSI victim

Government system, Madhya Pradesh, India (not a BFSI victim) — included for playbook relevance.

No MITRE list in the advisory summary. As described: stolen credentials, vulnerability exploitation, unsecured remote access for initial access; data theft + encryption + extortion; disabling security services, deleting backups, clearing event logs.

Advisory-level impact only for BFSI: reinforces Windows/AD hardening, backup protection and remote-access controls. No BFSI victim, ransom amount or data-loss figure was published in this item.

Critique: this is an advisory about a government victim in Madhya Pradesh. It is not a BFSI incident, and it should not be read as evidence of a new LockBit campaign against banks.

◐
No new in-window item found for FIN7 / Carbanak, Scattered Spider (UNC3944), Akira, RansomHub, Cl0p, or a FI-specific Initial Access Broker

No report found is not the same as no activity. It only shows what was published and searched in this window. The most recent Scattered Spider / FIN7 material in search results was older, so it is excluded from the 24-hour table. A 31 Oct report of a $25,000 offer of admin access to retailer Vijay Sales is a non-financial IAB example only, and proves nothing about banks.

03 — Vulnerability Watchlist

Patch, verify the build, then hunt

No new core-banking, SWIFT / payment-gateway or enterprise file-transfer CVE was reported in the 24-hour window in the sources searched. That is a limit of the search, not proof that none exists. The RBI item of 31 Oct is about API risk, not a new CVE. Where sources disagree on a score, both scores are shown — do not quote one without the other.

CVE IDCVSSAffected productExploitation statusRelevance to finance
CVE-2026-97229KEV due 1 Nov10.0GitLab CE/EE, self-managed
19.2 before 19.2.8; 19.3 before 19.3.5; 19.4 before 19.4.2. Root cause in Ruby / Onigmo regex engine.
● In the Wild — confirmed
KEV added 29 Oct. Alpine reported exploit attempts 20–21 Oct (8 worker crashes), rebuilt 27–28 Oct. Ransomware use: Unknown per CISA.
DevOps / CI platform holding source, secrets, runner and deploy tokens for banking/fintech pipelines. Custom-Ruby builds stay vulnerable after a GitLab upgrade — verify Ruby 3.3.13 / 3.4.12 / 4.0.9 or later as well as GitLab version. Fixed releases 19.2.8 / 19.3.5 / 19.4.2 (12 Oct) do NOT cover CVE-2026-110264 below.
CVE-2026-11026410.0GitLab CE/EE, self-managed — second regex RCE
Patched 21 Oct in 19.5.1, 19.4.3, 19.3.6, 19.2.9, 19.1.9, 19.0.10, 18.11.13
Not in KEV as of 31 Oct update
CISA record: Exploitation: none. No confirmed in-the-wild exploitation in sources read. Do not conflate with CVE-2026-97229.
Same pipeline risk. Upgrading only to the 12 Oct releases leaves this flaw open. Apply the 21 Oct releases.
CVE-2026-76418KEV due 1 Nov8.6Cisco Secure Firewall ASA & FTD
Remote Access SSL VPN, where a tunnel group uses secondary authentication
● In the Wild — confirmed
Cisco advisory 28 Oct: active exploitation in Oct 2026. KEV added 29 Oct. No workaround per Cisco.
Edge / VPN concentrators for remote banking staff and third-party access. Unauthenticated attacker forces device reload (DoS) — every VPN session drops. Secondary reporting cited >51,000 ASA SSL VPN devices internet-exposed (Shadowserver). Fixed: e.g. ASA 9.16.4.104, 9.18.4.96, 9.20.4.50, 9.22.3.200; FTD 7.0.11, 7.2.13, 7.4.9, 7.6.7.
CVE-2026-111357KEV due 2 Nov8.8 / 7.0
Discrepancy — see note
Ivanti Connect Secure VPN● In the Wild — as part of a chain
KEV added 30 Oct, forensic triage required.
VPN edge for privileged/remote access. Chained with CVE-2026-104935 (13 Oct) and CVE-2026-118527 to reach root; Volexity-tracked UTA0576 exploited the chain from 3 Oct against a fully patched appliance, targeting AD credentials. Fixes: 22.8R2.6 / 25.1.1.3 (29 Oct). 8.8 is Ivanti advisory score, 7.0 is CVE record — do not quote a single score without this caveat. A clean Integrity Checker does not rule out compromise.
CVE-2026-118527KEV due 13 Nov4.3Ivanti Connect Secure VPN● In the Wild — same chain
KEV added 30 Oct.
Predictable identifiers allow a remote authenticated attacker to upload attacker-controlled content to a known disk location — the file-write step in the root chain. Low standalone score understates chained impact.
CVE-2026-109618
CVE-2026-109619KEV due 29 Oct
9.2
Palo Alto severity*
PHP web servers; PAN-OS on PA-Series / VM-Series & Panorama (management interface) — “PHP2Shell”● In the Wild — confirmed
KEV added 26 Oct. Palo Alto advisory 21 Oct, updated 30 Oct: “Exploit Maturity: ATTACKED”, active exploitation against internet-exposed PAN-OS mgmt interfaces.
Web/payment and firewall management planes. Unauthenticated RCE. *9.2 is Palo Alto Networks severity for the PHP flaws on PAN-OS; CISA/PHP CVSS breakdown was not stated in sources read. Hotfixes include 12.2.4-h1, 11.2.4-h23, 11.1.4-h38, 10.2.7-h39. Context only: flaw family behind late-Oct Korean-bank/government warning (CYFIRMA, 24 Oct — outside window).
CVE-2026-41709KEV due 11 NovNot statedBroadcom VMware ESX (hypervisor)● In the Wild — per KEV listing
KEV added 28 Oct.
Virtualisation / cloud hosting for core and payment workloads: admin actions can bypass standard forensic logging — a logging blind spot that aids stealthy persistence and hinders IR. Critique: no CVSS was stated in the sources read, so severity cannot be judged from this briefing alone; “in the wild per KEV” also says nothing about how widespread exploitation is.
CVE-2026-46300KEV due 31 OctNot statedLinux kernel — socket-buffer coalescing during ESP (IPsec) input processing● In the Wild — per KEV listing
KEV added 28 Oct.
IPsec / VPN and host infrastructure: missing initialisation allows unintended modification of page-cache-backed data. Apply distribution kernel update and reboot.
CVE-2026-104935
Context — not new in window
Not statedIvanti Connect Secure● In the Wild — confirmed
Ivanti 13 Oct, KEV same day; auth bypass → unauthenticated command execution.
First link in the Ivanti chain above; patching it alone did not stop the Oct chain. Included so teams do not close the Ivanti item on this CVE alone.

CVSS gaps: No CVSS was stated in sources read for CVE-2026-41709, CVE-2026-46300 or CVE-2026-104935. Verify against NVD / vendor advisories before quoting in an external briefing.

04 — Supply Chain & Third-Party Risk

The compromise was in the vendor layer

Three different patterns in one window: a tampered RUM script, a subcontractor SaaS platform, and a hosted CRM claim. In each case the weak point was a supplier. In none of them is the full picture public yet.

Polymarket / DebugBear — script risk

~$2.19–2.71m

Estimates vary by researcher; Polymarket itself disclosed no total. ~700 wallets affected in reporting. Vendor not named by Polymarket; how the DebugBear script was tampered with is undetermined.

710Deposit wallets in analysis
731Addresses — ≠ victims
~1 hrIntl. trading halted

Lesson for BFSI: any third-party script on a customer-facing trading / payment page is production code. Inventory it, pin it, alert on change.

Critique: Polymarket has not named the vendor or explained how the script was changed. Until it does, other firms cannot check whether they run the same script — the lesson is clear, but the specific exposure is not.

🇯🇵 NTT Data — STdirectServices (31 Oct)

Service suspended after unauthorised access to an unnamed subcontractor’s SaaS platform. Per-firm counts — Rakuten up to 362 records, SBI Securities 1,169, MUFG Morgan Stanley 95, Kintetsu total reported 2,678 — overlap and are not additive. Subcontractor name, method and intrusion start date unknown. No login IDs, passwords, account numbers or trading details included, per Rakuten.

🇮🇳 AU Small Finance Bank + RBI API focus (31 Oct)

Bank investigating potential leak of limited vehicle-loan data on a Salesforce-hosted platform. Also on 31 Oct: RBI asked banks to examine APIs connecting them to non-bank lenders / fintechs, after CERT-In’s 9 Oct warning that bank API credentials were being stolen to move funds. Context reported 31 Oct: >700 unauthorised transfers, ~$2m loss across two Mumbai NBFCs via a compromised API sending loan-disbursement instructions to ICICI Bank (incidents 8–9 Oct).

Separately, The Statesman (31 Oct) reported an allegation concerning SBI customer data — see Sources. No technical IOCs were published for these exposures.

Critique: three separate things are easy to merge here and should not be. The AU claim is unverified. The RBI item is a regulatory warning, not an incident. The Mumbai NBFC losses are real reported incidents, but they date from 8–9 October, outside this window. None of them proves the others.

05 — Indicators & Behavioural Signals

What to hunt for — not just what to block

Take the FBI list from the FBI alert as republished by HackRead, not from a summary. Screen for derived and forwarded transactions, not just exact matches. A clean screen against this list does not mean funds are clean — it only means they do not match these 628 addresses.

FBI TraderTraitor list — 31 Oct 2026

628 addresses

Holding or having held stolen funds. Chainalysis warns the actor has already used thousands of addresses and expects tens of thousands — exact-match blocking alone will lag.

608Ethereum
19Bitcoin
1Solana

Behavioural pattern: rapid multi-chain sweep (~20 min), conversion predominantly into ETH, then DeFi / instant-swap laundering (THORChain, RPC nodes, instant exchanges).

FBI attribution via HackRead — full address list →

Polymarket / DebugBear

Malicious third-party RUM JavaScript reading logged-in trading tools / portfolio, selling positions or sending tokens to a hardcoded fallback address (Arkham-linked wallet holding ~$2.65m). Watch for: any third-party script change on a trading/payment page, a production build note removing a vendor script (Polymarket build timestamped 17:59 UTC), and unexpected position sales with no credential-theft artefact.

Ivanti chain

Unexpected files on a Connect Secure appliance, admin logins and AD credential access, exploitation from 3 Oct against fully patched appliances (UTA0576). A passing Integrity Checker is NOT a clean bill of health.

GitLab

CI/CD config changes from newly created / untrusted accounts, worker crashes (Alpine: 8 crashes over 20–21 Oct), and downloads of the project’s container image by an unknown party (exploit tuning). Check both GitLab version AND bundled Ruby version.

Cisco ASA / FTD

Unexpected firewall reloads (show crashinfo) on SSL VPN devices with secondary authentication in any tunnel group, including DefaultWEBVPNGroup.

PAN-OS / PHP

Internet-reachable management interfaces, admin / config-log changes, PHP hosts with unpatched branches — treat any internet-exposed PAN-OS management interface as possibly compromised even after hotfixing.

Social indicators — NTT / AU / SBI exposures

No technical IOCs published. The indicator is social: phishing / vishing / smishing quoting a real bond name, purchase amount (e.g. ¥1m+ Kintetsu bond), loan amount, vehicle or EMI date, CIF ID or branch. Any message about “perk”, “compensation”, KYC update or “security check” referencing those details should be treated as suspect even if the caller knows correct personal data.

Critique: this warning rests on the firms’ statements about which data was exposed. If more data was taken than stated, the same messages become more convincing, not less. Correct personal details are not proof that a caller is genuine.

Butterfield vishing pattern

Two-stage calls (fake hack alert, then fake “expert”), fraudulent live support-chat link, requests for device access — including a US (Vermont-appearing) caller number in one victim report.

Payment-diversion — screen these beneficiaries

ZZDJB CO LIMITED — HSBC Hong Kong
ZHONGXIAN TRADING PTE. LTD — UOB Singapore

Plus: lookalike-domain email, WhatsApp / personal-email payment instructions, and senior-executive / lawyer impersonation without any malware or mailbox-compromise artefact (Capillary pattern, KPMG findings 31 Oct).

India API fraud pattern

Anomalous loan-disbursement API volumes, new / changed beneficiary accounts, and API credential use outside normal patterns.

Also in-window context: Avast AI-phishing warning (updated 31 Oct, via Particle News) reusing data from mid-Oct PHP hacks — AI-scaled phishing, not a new trojan disclosure. No mobile-banking trojan was newly reported on 31 Oct–1 Nov in sources searched (Hirudo reporting is 14–15 Oct and is excluded).

06 — Defensive Actions

Prioritised — in order

Two queues: SOC hunts and blocks today; Vulnerability Management closes each flaw and proves closure. “Patched” is a claim too — it should be checked, not just recorded.

🛡 SOC — do first

  1. Ingest and block the FBI TraderTraitor list today628 addresses + derived-transaction screening; alert on ETH-conversion / DeFi-swap patterns; brief crypto/fraud desks that exact-match lists will lag address rotation.
  2. Hunt, don’t just patch — Ivanti, GitLab, CiscoReview Ivanti admin logins/files and AD credential use; GitLab CI changes by new accounts, worker crashes and token use; Cisco unexpected reloads. Rotate runner/deploy tokens and privileged credentials where any sign of exploitation is found.
  3. Freeze third-party scripts on customer-facing finance pagesInventory all loaded vendor scripts (the DebugBear lesson), enforce subresource integrity / change approval, alert on script-content change or unexpected authenticated-session transactions with no login anomaly.
  4. Raise phishing / vishing defences for NTT / AU / SBI populationsWarn customers and front-line staff callers may quote real bonds, loans, CIF IDs and branches; enforce callback-to-known-number, never-ask-for-OTP/PIN/password rules; monitor lookalike domains and WhatsApp payment instructions.
  5. Treat exposed PAN-OS mgmt & PHP hosts as assumed-compromisedUntil log review is complete; restrict management to a jump host with permitted-IP allowlisting only.
  6. Lock down privileged endpoints that can reach signing / withdrawal systemsDevOps laptops, service containers and hot-wallet signing infrastructure need phishing-resistant MFA, EDR, and separation so one laptop cannot submit unauthorised withdrawals; rehearse withdrawal-freeze and stablecoin-freeze playbooks.

🔧 Vulnerability Management

  1. TODAY — 1 Nov deadline already reachedConfirm Cisco CVE-2026-76418 and GitLab CVE-2026-97229 are closed — including secondary-auth tunnel groups on every ASA/FTD and Ruby version on every self-managed GitLab — or take the service out of internet reach until patched.
  2. By 2 Nov / 13 Nov — IvantiCVE-2026-111357 to 22.8R2.6 / 25.1.1.3 by 2 Nov; CVE-2026-118527 by 13 Nov. Do not close on version alone if exposed during the Oct chain — rebuild suspect appliances.
  3. Immediately afterApply 21 Oct GitLab releases for CVE-2026-110264, PAN-OS PHP hotfixes, VMware ESX logging fix (due 11 Nov) and Linux kernel updates (due 31 Oct).
  4. Third-party / fourth-party attestationsDemand incident attestations from Salesforce, NTT Data and any subcontractor/SaaS in the chain; map sub-processors two levels down for any service holding customer PII or payment instructions; add script-vendor and API-credential controls to vendor risk assessments (RBI API focus).
  5. API security — India lesson, applicable generallyRotate and scope bank/fintech API credentials, enforce mutual authentication and per-counterparty limits, alert on disbursement-API volume / beneficiary anomalies.

Fraud & attack-vector watch — honest scope

Fraud reporting in this window centred on vishing, payment diversion, and API-credential abuse. It did not include a newly reported mobile banking trojan, infostealer family, or DDoS-extortion campaign against a bank. Account-takeover risk here sits downstream of exposed KYC / bond / loan data (AU, NTT) and of authenticated-session abuse (Polymarket). It is not a confirmed credential-stuffing wave. Keep monitoring, but do not re-label these incidents as proof of account takeover.

07 — Open Questions

Could not verify

  • ScarlettGroup / AU SFB: The breach itself is unconfirmed. Salesforce denies a platform vulnerability. The affected count, the data volume and the way in are all undisclosed. Do not report this as a confirmed Salesforce compromise — and do not report Salesforce’s denial as proof that nothing happened.
  • NTT Data: The subcontractor’s name, the attack method and the intrusion start date are unknown. The overlap between the broker counts (362 / 1,169 / 95 / 2,678) is also unknown. The counts are per firm and per holding. They are not one total and must not be added.
  • Binance: The total loss is reported as $6bn, $6.3bn and $6.4bn by different sources. The frozen figure (~$2.7–3bn) and the ~$2.4bn still held are estimates from researchers and partners, not a settled FBI figure. How the DevOps laptop was compromised is not publicly explained. Neither are the “additional breach points” Binance mentioned.
  • Polymarket: The loss total ($2.19m vs $2.65m vs $2.71m) and the victim count are researcher estimates. Polymarket has not named the vendor and has not published its own total. How the DebugBear script was tampered with is still undetermined.
  • Not found in this window: new FIN7 / Scattered Spider / Carbanak / Akira / RansomHub / Cl0p attacks on BFSI victims; FI-specific IAB sales; SWIFT / core-banking / file-transfer CVEs; DDoS extortion against banks; or a mobile-banking trojan newly reported on 31 Oct–1 Nov. This is a limit of what was published and searched. It is not evidence that these actors were inactive.
  • CVSS gaps: No CVSS was stated in the sources read for CVE-2026-41709, CVE-2026-46300 or CVE-2026-104935, so their severity cannot be judged from this briefing. Ivanti CVE-2026-111357 has conflicting scores (8.8 vs 7.0). The PHP 9.2 figure is Palo Alto’s severity rating, not a CISA breakdown.
08 — Sources

Reporting dated in the window

These are the published sources used, read as articles — not checked inside vendor portals. Dates are the dates shown by each source. A link shows where a claim came from. It does not by itself prove the claim.

How to use this briefing: “Confirmed” and “unverified” call for different actions. Block and hunt on confirmed items today. On unverified claims, monitor and prepare customer warnings, but do not state them as breaches outside the bank. Check CVSS scores and fixed versions against NVD and vendor advisories before quoting them outside the SOC. Research date: Sunday 1 November 2026, SGT (evening).
← Back to main page