Microsoft Security Update Guide (September 2026; MSRC releaseDate 17 September 2026 PDT) lists CVE-2026-85887, an M365 Copilot information-disclosure vulnerability: incorrect permission assignment for a critical resource (CWE-732) lets an authorized (low-privilege) attacker disclose information over the network. Microsoft CVSS 3.1 base 7.7 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:N); MSRC severity Critical / impact Information Disclosure. Exclusively hosted service: Microsoft states the issue is already fully mitigated — no customer patch steps; CVE published for cloud transparency. MSRC: publicly disclosed No; exploited No. NVD received 18 September 2026 00:17 UTC. Distinct from Azure AI Foundry CVE-2026-85889/85917. (sources: MSRC; NVD)
MSRC — CVE-2026-85887 M365 Copilot information disclosure (17 Sep 2026)
AIR Security (Or Nevo, Dor Granat, Niv Hoffman; 17 September 2026) discloses Plugin4Shell: a marketplace plugin SHA-pinning bypass affecting the four major AI coding agents — Anthropic Claude Code, OpenAI Codex, GitHub Copilot, and Google Gemini CLI. Attack path: a trusted plugin/skill repo is compromised (building on prior SkillJacking findings); the agent checks out the pinned commit but does not verify the tree that landed, so checkout can resolve to malicious code while the pin still appears honoured — zero-click RCE on the developer host with the employee’s full reach into enterprise systems. No CVE identifier published in the disclosure. Patches named by AIR: Claude Code 2.1.179 (Anthropic); Codex 0.146.0 (OpenAI). GitHub Copilot: disclosed to Microsoft, no patch shipped at publication — users have no vendor fix yet. Gemini CLI: Google deprecated the CLI and will not patch; AIR advises migrating to Antigravity (no marketplace SHA-pinning surface). Enterprises using Air Marketplace / Air Filter were not affected per the authors. Treat community agent plugins as untrusted code until agents verify checkout integrity.
CVE-2026-92838 (published ~17 September 2026 per Tenable) is a DLL hijacking issue in the GeoVision GV-Remote E-Map desktop application: one or more DLLs are loaded from an unsafe search path. A local attacker with write access to a directory searched before the legitimate library location can plant a malicious DLL and achieve code execution as the GV-Remote E-Map process. Tenable CVSS 3.1 base 7.8 (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H). No public in-the-wild exploitation claim on the Tenable snippet reviewed. Relevance: physical-security / VMS operators running GeoVision Windows clients.
CBS News (16 September 2026, 7:45 pm US Eastern; 17 September, 7:45 am Perth) reports, citing US officials, that US Coast Guard personnel and FBI agents boarded two Texas-bound energy tankers last month after cyberattacks struck the vessels while they were travelling toward the United States. One was the VL Prosperity, a 1,093-foot Liberian-flagged crude oil tanker; public vessel data cited by CBS shows it left Egypt's Sidi Kerir oil terminal on 1 August for Galveston, and US officials say it slowed near the Strait of Gibraltar around the time of the attack. Iran's Mehr News Agency reported on 20 August that the ship was attacked on 7 August in the Strait of Gibraltar and, citing an unnamed crew member, alleged hackers breached the engine room, reducing engine cooling flow, increasing engine speed and interfering with fuel systems; Iranian state media also claimed access to its propulsion, navigation and cargo systems and a 30-hour communications outage. The Coast Guard has not publicly attributed the attacks to Iran. On 21 August, the day after Mehr's report, Coast Guard cyber personnel, law enforcement officers, a vessel inspector and FBI Cyber Action Team operators boarded the ship for four days. Rear Adm. Amy Grable, commander of US Coast Guard Cyber Command, told CBS investigators found malicious cyber activity on the vessel's IT and other onboard systems but no evidence the ship had become unsafe to navigate; the boarding was one of roughly 40 to 50 Coast Guard Cyber Protection Team missions in the past year. The US is investigating whether the two attacks are connected and whether Iran or another foreign adversary was behind them. OT/maritime relevance for AU shippers and ports. (source: CBS News) Correction (26 Oct 2026): the card now rests on CBS News's own report rather than SecurityWeek's summary of it, and a Wall Street Journal detail on when the second ship was boarded, which no source readable to Cyberstack supports, has been removed. (source: CBS News)
CBS News — Coast Guard and FBI boarded two energy tankers after cyberattacks (16 Sep 2026 US Eastern; 17 Sep Perth)
CVE-2026-81546 (published ~17 September 2026 per Tenable/NVD indexing) covers a stack-based buffer overflow in the Affinity by Canva application before the 3.3.0 September 2026 release: inadequate bounds checking when parsing Affinity document files. A crafted Affinity document opened by a user can lead to arbitrary code execution. Tenable lists CVSS 3.1 base score 7.7 (High). Australia relevance: Affinity is Canva’s creative suite (Canva is Australian-headquartered). Patch: upgrade Affinity by Canva to 3.3.0 or later. No public exploitation claim on the Tenable/NVD snippets reviewed this pass. (source: CVE/NVD/Tenable record)
Tenable — CVE-2026-81546 Affinity by Canva (indexed 17 Sep 2026)
Doyensec research post (17 September 2026) shows Open vSwitch’s kernel datapath can strip SKBFL_SHARED_FRAG from a still-forwarded packet, re-opening the Fragnesia Dirty COW-class primitive: an unprivileged user can cause in-place ESP decrypt over page-cache pages they may only read, writing attacker-chosen bytes into root-owned file page cache. Tracked as CVE-2026-90049, CVE-2026-89487 and CVE-2026-80977; reported to the Linux kernel security team and coordinated with OVS maintainers. Builds on prior Dirty Frag / Fragnesia work (including CVE-2026-43284 and CVE-2026-43500). Impact surface: virtualisation/container stacks using OVS (OpenStack Neutron, oVirt, Antrea/OVN-Kubernetes, libvirt bridges, etc.). (source: Doyensec blog)
BleepingComputer (17 September 2026) reports the US FBI seized nightmare-stresser[.]com and nightmarestresser[.]org used by NightmareStresser, a long-running DDoS-for-hire (booter) service. FBI Cyber Division said that since 2022 the service was used to launch hundreds of thousands of actual or attempted DDoS attacks worldwide. Seizure banners cite Operation PowerOFF, the international law-enforcement effort against DDoS-as-a-service infrastructure. Historical context: Searchlight Cyber (2023) previously assessed ~566k registered users and up to ~200 Gbps multi-layer attacks; DOJ had seized nightmarestresser[.]com once before in December 2022 with related arrests. (source: BleepingComputer quoting FBI Cyber Division)
ESET Research (17 September 2026; “Beware the SparroWock”) documents SparroWocky, the new flagship modular C++ backdoor of China-aligned APT FamousSparrow, replacing SparrowDoor in campaigns focused on Latin American government targets (Argentina, Ecuador, Guatemala, Honduras, Panama, Peru, Puerto Rico, Venezuela per BleepingComputer). ESET attributes with high confidence (early SparroWocky deployments via SparrowDoor). Capabilities include command/file execution, in-memory BOF load, system/network/user enumeration, file ops, screenshot streaming, cross-session process create, TCP proxy, and self-removal; persistence via Windows service (e.g. ProcAuditManager) or registry Run key (e.g. SnapCart); DLL side-load after RC4 .dat decrypt. Evasion includes MinHook CreateThread spoofing (AnimateWindow start address), call-stack spoofing, and dynamic API resolve. C2 over 443/8080 or HTTP/SOCKS5 proxies (ESET lists ≥18 addresses). No CVE. (sources: ESET Research; BleepingComputer)
ESET Research — Beware the SparroWock (17 Sep 2026)
CISA published guidance (September 2026; SecurityWeek 17 September) on deploying cyber decoys to strengthen detection and response for critical infrastructure. Decoys complement Zero Trust by assuming breach and helping organisations detect, observe, and block malicious activity. Document: Using Cyber Decoys to Strengthen Detection and Response (508c PDF). No CVE. (sources: CISA PDF; SecurityWeek)
CISA — Using cyber decoys to strengthen detection and response (Sep 2026 PDF)
Two CVE records published on 16 September 2026 (AWST) rate CVE-2026-71133 and CVE-2026-83059 at CVSS 3.1 10.0, critical. CVE-2026-71133 is in the Authentication Engine component of Oracle Access Manager (Oracle Fusion Middleware): the record says the easily exploitable vulnerability allows an unauthenticated attacker with network access via HTTP to compromise Oracle Access Manager, and a successful attack can result in takeover of Oracle Access Manager. CVE-2026-83059 is in the OID LDAP Server component of Oracle Internet Directory: the same kind of unauthenticated network attack, via LDAP, can result in takeover of Oracle Internet Directory. For both, the records say attacks may significantly impact additional products (scope change), and list supported versions 12.2.1.4.0 and 14.1.2.1.0 as affected. (source: CVE records) Correction (31 Oct 2026): removed Oracle's Critical Security Patch Update advisory and SecurityWeek, which this desk cannot read, as sources, with the details only they gave; the card now carries only what the two CVE records state, and its date moved from 15 to 16 September 2026, the records' publication date in AWST. (source: CVE records)
The US Department of Health and Human Services (HHS) Office for Civil Rights breach portal lists AngMar Management Services, a Texas business associate, with 126,196 individuals affected, in a report submitted on 16 September 2026 as a hacking/IT incident involving a network server. The portal entry does not describe what information was involved. (source: HHS OCR breach portal) Correction (25 Oct 2026): the card now rests on the HHS breach portal entry; details from a press report we cannot verify (the type of data, the timeline and a leak-site claim) have been removed. (source: HHS) Correction (25 Oct 2026): the card date has moved from 5 October to 16 September 2026, the breach report submission date the HHS portal lists; HHS gives a US date with no time. (source: HHS)
HHS Office for Civil Rights — breach portal: cases under investigation (AngMar Management Services, TX, 126,196, submitted 16 Sep 2026)
Trend Micro Zero Day Initiative advisory ZDI-26-708 / ZDI-CAN-29849 (published 16 September 2026; CVSS 5.3 AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L) documents an unpatched Microsoft Windows privilege-escalation issue in HTTP proxy authentication handling. A low-privileged local attacker who can change the user proxy setting to an attacker-controlled server can coerce NTLM authentication in the machine-account context (commonly via Network Connectivity Status Indicator / WinHTTP connectivity checks running as NETWORK SERVICE). Microsoft acknowledged the report (May 2026), initially indicated a September fix, then on 31 August 2026 said the issue did not meet the Windows security servicing bar; ZDI published as a 0-day on 16 Sep. No CVE assigned. ITRESIT Labs (28 Sep) describes a pentest chain using the coercion into AD CS Web Enrollment relay when EPA/HTTPS are not enforced — that chain is environmental, not part of ZDI’s scored impact. (sources: ZDI-26-708; ITRESIT)
ZDI-26-708 — Microsoft Windows HTTP Proxy Privilege Escalation (0-day, 16 Sep 2026)
