Cyber Digest
A daily roundup of key cybersecurity developments across sectors
Executive Summary
A supply-chain compromise of ConnectWise's Cloud Backup (formerly SkyKick) service is the weekend's dominant story, with attackers abusing a stolen application identity to sign into more than 170 customer Microsoft 365 tenants before the vendor disabled seven application identities and suspended the service — the third MSP/cloud-channel compromise this week after AOMEI's update servers and GhostAction's CI workflow takedowns. In exploited-software news, a critical OS command-injection flaw in the SiteGuard web application firewall (CVE-2026-106931) is being exploited in the wild with root/SYSTEM privileges and is now on Japan's JPCERT/CC radar, while Drupal shipped urgent patches for two critical remote code execution flaws it warns could be weaponised within days. On the security-industry side, Proofpoint announced a US$6.3 billion all-cash acquisition of Varonis, and Bank of America revealed it hired 1,000 contractors to remediate vulnerabilities surfaced by Anthropic's Mythos model at a cost of "a couple of hundred million" dollars.
There is no direct Australian incident in this edition, but the ConnectWise incident carries direct weight for local managed service providers: SkyKick/ConnectWise Cloud Backup is widely deployed by Australian MSPs to back up M365 and Google Workspace tenants, and the attack pattern — a stolen service-principal credential authenticating as the application, bypassing MFA entirely — is exactly the application-identity risk APRA and the ASD have flagged in cloud-guidance updates. Sonic Healthcare, the ASX-listed pathology group, is drawn into the US AMCA litigation settlement through its Sonic Healthcare USA subsidiary, one of four defendants paying US$6.4 million. Drupal's RCE patches matter for the many Australian government, university and non-profit websites running the CMS. On the ACSC front — all three listings re-checked 19 October — there is no new alert or advisory; the most recent output remains the 14 October alert *Protecting operational technology and edge devices* (covered here for the first time but outside the freshness window), following the 9 October advisory on Chinese state-linked actors combining automated and hands-on hacking tools. ACSC's ASD Essentials chaptered guidance (15 October news item) was covered in the 17 October edition.
No new NZ privacy-breach notifications this week: the Privacy Commissioner's site, checked 19 October, shows no media release newer than the Compliance Notices issued to Manage My Health and Health NZ (23 September, stemming from the Commissioner's Phase 1 report into the late-2024 Manage My Health cyber incident) and the Commissioner's expectations statement on smart glasses. NZ Police news (checked 19 October) has no cyber content — the newest releases cover an avalanche fatality at Mount Dixon (18 October) and gang/drug arrests in Gisborne (16 October). CERT NZ remains unreachable to automated checks; Pulse+IT's New Zealand digital-health coverage this weekend (ACC digital notes, Valentia hiring) carries no cyber dimension.
In the week to 18 October (12–18 October), the digest corpus carried 57 stories: Zero-day/Vulnerability led with 16, followed by Malware (15), Breach/Data Leak (10) and Ransomware (8); geography was US-heavy (26) with Australia second (21) and APAC third (4). The week's through-line is the weaponisation of trusted intermediary channels — AOMEI Backupper update servers, Shopify checkout extensions, GhostAction CI workflows and now ConnectWise's backup platform — which reads as a deliberate shift toward supply-chain and managed-channel targets where a single credential grants attacker-scale access to many victims. Running second is the edge-software patching wave (BeyondTrust, PAN-OS, Cisco SEG, FortiOS/PEERSTITCH, SiteGuard, Ivanti) now visible in both CISA KEV additions and national-CERT alerts, including two KEV additions on 16 October. Two AI signals converged this weekend: Cisco Talos detailed UAT-11985, an AI-assisted spear-phishing operation against Taiwan research institutions, and Bank of America quantified the remediation cost of AI-discovered vulnerabilities — expect AI-generated phishing volume and AI-assisted patch reverse-engineering (Sansec's warning on BlockStrap) to keep featuring through the week ahead.
Incident Map
Global (Macro) 4 stories
ConnectWise suspends Cloud Backup after attackers compromise more than 170 Microsoft 365 tenants
Unknown threat actors abused a compromised application identity belonging to ConnectWise Cloud Backup (formerly SkyKick) to authenticate directly into customer Microsoft 365 tenants; Huntress observed confirmed unauthorized sign-ins in 173 partner/customer tenants using the stolen credential, which works against every tenant where the backup app was installed. Microsoft confirmed unauthorized authentication attempts from a single IP (144.172.64[.]128) beginning 15 October; Huntress says the actor moved through tenants in sorted GUID order, pointing to a scripted target list, and in one tenant a separate ConnectWise app granted itself Global Administrator before modifying other trusted applications to establish persistence. Because the attacker authenticated as the application itself, MFA did not apply. ConnectWise identified the breach on 14 October and on 17 October disabled seven application identities and revoked certificates, suspending Cloud Backup; it says archives remain encrypted and shows no evidence they were accessed. The incident is the week's third MSP/cloud-channel compromise and leaves every customer running the backup app to audit Entra ID permissions for rogue service principals.
Critical SiteGuard WAF command-injection flaw (CVE-2026-106931) exploited in the wild
Threat actors are exploiting an OS command-injection vulnerability in the statistics feature of SiteGuard Server Edition web application firewall instances, executing commands with root or SYSTEM privileges. Tracked as CVE-2026-106931 and rated critical by the vendor and Japan's JPCERT/CC, the flaw affects SiteGuard Server Edition for Apache, Nginx and IIS versions 2.20 to 8.00 Update 4 (and the FreeBSD variants before 7.00 Update 6); the statistics function is enabled by default. SiteGuard claims protection of 1.5 million websites, with the plugin active on roughly 500,000 WordPress sites. The vendor warns that remediation may erase attack traces and advises preserving diagnostic evidence before patching; customers with direct support contacts were notified on 29 September. Neither advisory names the attackers, victims or scale.
Drupal patches two critical RCE flaws, warns exploits may emerge within days
Drupal has released fixes for two critical remote code execution vulnerabilities affecting all supported versions of the content management system: CVE-2026-108097 in core's Field UI module and CVE-2026-108099 in the contributed Entity List module, both exposing at least one unauthenticated exploitation path when combined with contributed modules or custom code. The Drupal Security Team pre-announced the coordinated release (PSA-2026-10-12), warning that "exploits might be developed within hours or days" and urging administrators to reserve time for the update window. Administrators must update core to 10.6.19, 11.3.19 or 11.4.10 depending on branch, plus Entity List 3.0.10. Drupal runs more than one million websites, including many government, university and non-profit properties, making this patch wave a priority for public-sector teams before exploit code circulates.
MALFEX campaign continues: malicious npm package 'imgdecode' hides shellcode loader in native addon
Researchers at SafeDep identified a malicious npm package, "imgdecode", that conceals a loader inside a compiled Node.js addon: a nine-line JavaScript file loads a Windows DLL which downloads encrypted shellcode from GitHub and injects it into a new dllhost.exe process when require('imgdecode') is called from a Windows console. Uploaded 30 September by a publisher named "malfexteste8" and downloaded 979 times (still available at time of writing), it is the latest in the MALFEX campaign — an operation active since August 2023 across five npm publisher handles that has used Overlord RAT loaders and Node.js stealer chains, and which has repeatedly re-published replacement packages within days of npm removals. The ongoing cat-and-mouse confirms the need for registry-level package vetting as part of the software supply chain.
Indicators of compromise · Overlord RAT — 4 shown
hxxps://fileserver[.]devin-hk[.]shop/api/file-share/6300dae9-a167-418e-9f6c-db02e0f005f4/downloadhxxps://github[.]com/mariaaefyw/fileserver/releases/download/v1005/fileserver-update[.]datb2aa2c5c202e530bb1f8ed9f1c00b4e59be8445455f7eef1b8fb25a2f92408a6expirience[.]icu
Defanged third-party indicators (abuse.ch). The defanging is deliberate: never click, resolve or fetch these values. An indicator corroborates a report — it never proves one, and its presence here does not mean this story's hosts are listed.
Government 1 story
Anne Arundel County confirms August breach as Rhysida's 2.6TB extortion claim widens
Officials in Anne Arundel County, Maryland — a Baltimore suburb of more than 600,000 residents that includes the state capital Annapolis — confirmed that hackers stole files from county systems during an August incident, ending months of silence after the Rhysida ransomware gang claimed the attack and threatened to leak 2.6 terabytes of documents unless a US$2 million ransom was paid. The county says it found "irregular activity" on 28 August and determined an unauthorized actor accessed "a limited number of systems" and copied files; investigators are still assessing whether resident data was taken. County officials notified federal law enforcement and other agencies. Rhysida has been active on government targets, recently attempting to extort Berlin's government after stealing emails and contracts.
Legal Services 1 story
US$6.4 million settlement agreed in AMCA customer data breach litigation
Plaintiffs and four defendants — Sonic Healthcare USA, Aurora Diagnostics, Clinical Pathology Laboratories and Austin Pathology Associates — have agreed to pay US$6,438,578.20 to resolve the "Other Labs Track" of the sprawling litigation arising from the 2018–2019 cyberattack on American Collection Medical Agency (AMCA), the medical-debt collections firm whose web payment portal compromise exposed data of more than 27 million individuals. Consolidated in the US District Court for the District of New Jersey after six years of litigation, the settlement resolves claims for negligence and breach of contract while the defendants continue to deny liability; the main track against AMCA itself is unaffected. For Australian readers the case matters twice over: Sonic Healthcare USA is the US arm of ASX-listed Sonic Healthcare, and the six-year litigation tail shows the longevity of breach-related legal risk.
Financial Services 1 story
Bank of America hired 1,000 contractors to patch flaws found by Anthropic's Mythos model
Bank of America CEO Brian Moynihan disclosed that the bank brought in 1,000 contractors, on top of 3,000 internal staff, to remediate software vulnerabilities uncovered by Anthropic's Mythos AI model earlier this year — work that contributed to a "couple of hundred million plus" dollars in second-quarter spending alongside AML compliance costs. Speaking to CNBC after the bank's third-quarter earnings, Moynihan said the contractors "remain here today" and that the patching pipeline should deliver declining costs over time. The disclosure is the first quantified picture of what banks and regulators worldwide have called the hardest part of the Mythos fallout: triaging and fixing a flood of AI-discovered flaws across legacy platforms. For financial-sector CISOs it is a concrete data point on the cost curve of AI-assisted vulnerability discovery — and a preview of what APRA-regulated entities may face as AI auditing tools mature.
Defence 1 story
UAT-11985: AI-assisted phishing targets Taiwan research affiliates with fake event invites
Cisco Talos has detailed a spear-phishing operation, tracked as UAT-11985, targeting individuals affiliated with Taiwan research organisations with adversary-in-the-middle (AitM) phishing designed to harvest Google account credentials and MFA codes. Observed since mid-2026, the campaign impersonates legitimate institutions — the Taiwan European Union Centre, NCCU Institute of International Relations and Taiwan Research Institute — and builds lures from real public event details, with near-identical message structures across targets that Talos says point to AI-assisted mass production from a reusable template. The actor also modified legitimate event posters with malicious QR codes (quishing), extending the attack surface beyond email to printed or shared materials. The campaign fits the pattern of espionage-focused credential harvesting against Taiwan's research and policy community ahead of regional flashpoints.
Retail & Entertainment & Sport 1 story
Critical BlockStrap flaw lets hackers run code on Magento and Adobe Commerce servers
E-commerce security firm Sansec has warned of a maximum-severity vulnerability (CVE-2026-103849), dubbed BlockStrap, affecting Magento Open Source and Adobe Commerce through 2.4.9 and Adobe Commerce B2B through 1.5.3 — a second-pass template-engine directive injection in child email templates that lets attacker-controlled content execute arbitrary PHP code on store servers. Adobe rated the flaw "priority 2" (patch within 30 days) in bulletin APSB26-153, but Sansec disputes that assessment, noting it identified four attack paths within 24 hours and that AI agents can reverse-engineer the patch into working exploits quickly — the same complexity profile as StyleSmuggler, the Magento email-template zero-day that went from hotfix to mass exploitation in three days. No in-the-wild exploitation has been observed yet, but Sansec expects attack waves soon and warns merchants need both the StyleSmuggler hotfix and the 2026-oct release.
Transport 1 story
California trucking company owner pleads guilty in $2.1 million identity-theft scheme
Hasan Korkmaz, 45, the owner of San Diego Logistics Group Inc (Santee, California) and a Las Vegas resident, pleaded guilty on 15 October to all 18 counts of bank fraud, money laundering and aggravated identity theft in a scheme that used nearly 100 stolen identities — names, birth dates, Social Security numbers and more — to open fraudulent credit card accounts in a "bust-out" pattern, running up more than US$2.1 million in losses across two banks. Prosecutors traced proceeds to accounts he controlled in Turkey; Korkmaz fled the US for roughly 18 months before being arrested on his return in June 2025. The case, pursued by the US Attorney's Office for the Southern District of California, is a reminder for the transport and logistics sector that identity theft and synthetic-fraud vectors hit carriers through back-office and credit processes, not only through operational systems.
Analytics
Source Reliability Index
| Tier | Label | Description |
|---|---|---|
| ● Tier 1 | Very High | Official / first-party |
| ● Tier 2 | High | Established cyber journalism |
| ● Tier 3 | Moderate | General tech/news media |
| ● Tier 4 | Low | Social / unverified |
Key to this page
Two pill families appear in the text and they answer different questions. A CVE pill colours severity — a measured CVSS band from the National Vulnerability Database. A threat-actor pill colours attribution confidence — how well-corroborated the naming is, which is a claim rather than a measurement. Both are links: a CVE opens the ATT&CK matrix or its wiki page, an actor opens its wiki page.
CVE identifiers
- CVE-XXXX-NNNNCritical · CVSS 9.0+
- CVE-XXXX-NNNNHigh · CVSS 7.0–8.9
- CVE-XXXX-NNNNMedium · CVSS 4.0–6.9
- CVE-XXXX-NNNNLow · below 4.0
- CVE-XXXX-NNNNNo severity resolved — not the same as low
Threat actors · MITRE ATT&CK
- APT29State attribution stated by MITRE ATT&CK
- ShinyHuntersSelf-declared, or criminal-reporting attribution
- Transparent TribeContested — ATT&CK hedges, or two plausible sponsors
- ZIRCONIUMNo attribution in MITRE ATT&CK
Story signals
- ● Tier 1/4Source reliability — 1 official, 4 leads only
- VerifiedCorroborated by a second source or the principal
- ReportedSingle outlet, or a claim still in progress
- UnverifiedA claim we could not corroborate
- ConfirmedBreach acknowledged by the victim or a regulator
- ProbableBreach indicated but not yet acknowledged
- IOCs · FamilyLive abuse.ch indicators exist for that malware family
A collapsed Indicators of compromise block under a story lists defanged abuse.ch indicator values. The defanging is deliberate — never click, resolve or fetch them. An indicator corroborates a report; it never proves one.
Full methodology, evidence grading and caveats: Methodology & reading guide →