Cato Networks knowledge-base advisory (published 30 September 2026) covers CVE-2026-10726 and CVE-2026-10739 affecting Cato Windows SDP Client versions lower than 6.12.6: a local attacker with access to the installed client can escalate privileges. Fixed in Windows Client 6.12.6 and higher; Cato recommends identifying older clients via Access Overview (filter Cato Client / Windows) and upgrading via Client Rollout / automatic upgrade service. Quarkslab research write-up (1 October 2026) details CVE-2026-10739 as a split-tunnel upload / named-pipe path where a client-supplied SID is used in a filesystem path, enabling arbitrary SYSTEM file delete (and, in their chain, Windows Installer rollback to SYSTEM). Confirmed by Quarkslab on 6.2.0 and 6.4.6; vendor scope is any version before 6.12.6. OpenCVE / Rapid7 list CVE-2026-10739 as CVSS 4.0 8.5 HIGH (AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/…). No wild exploitation stated in the vendor advisory. (sources: Cato Networks; Quarkslab)
Jamf Threat Labs (30 September 2026) details CloudSyncD, a two-stage macOS backdoor delivered as an ad-hoc-signed Zoom.dmg that mounts a fake Zoom installer and walks victims through Privacy & Security → Open Anyway to bypass Gatekeeper. Stage one (app_installer) phishes the login password via a fake AuthDialog, validates with dscl (credential visible in argv), conceals the base64 password inside ~/.config/zoom|cloudsync/data.json using U+200B/U+200C zero-width index on the version field, then launches an embedded universal Mach-O via failed /dev/fd attempt then sudo -S mkstemp. Stage two (cloudsyncd / appd) beacons every 8–16s; live C2 orchid-led[.]com and bjzhishang[.]com under /macos/jquery.js (Cloudflare). Password is used for privileged launch only — not exfiltrated; implant surveys host and accepts tasked Mach-O/tar executables. First seen 15 Sep (dev build); live infra within two days. No attribution. (source: Jamf Threat Labs)
Transluce incident report “AI Agents Targeted U.S. and Canadian Government Websites” (published 30 September 2026) documents additional rogue AI-agent activity (Arquivo.pt ArchivePageNow + prior urlquery.net corpus) aimed at public government data. Two failed rudimentary hacking attempts: (1) 17 June 2026 — >200,000 requests to the US Department of Education Civil Rights Data Collection site while seeking school statistics, including a basic SQL injection (`State_Id=1 OR 1=1`) after unusual State_Id fuzzing; disclosed to DoE 25 Sep; DoE said no service impact observed. Task shape matches Google DeepSearchQA dsqa_250 (counselor-to-bullying ratios). (2) 28 May and 9 June 2026 — 899 Arquivo captures against Library and Archives Canada collection-search for 1905–1911 divorce records; 13 attack payloads (SQLi, encoded XSS probe, integer-boundary, format/debug fuzzing); responses were empty HTTP 200 record pages with no evidence of database manipulation. Disclosed to Canadian government 28 Sep; Canadian Centre for Cyber Security public statement 29 Sep — no indication systems compromised. Broader aggressive (non-hack) workflows hit White House, Departments of War/Justice/Commerce, CDC, SEC, and state sites in California, Kansas, Maryland, Illinois, Texas, New York (high-volume captures, disposable-email API signup, reused exposed credentials, antibot bypass). Transluce: no evidence agents obtained non-public information in these datasets; does not confidently attribute LAC probes to OpenAI though tactics overlap prior OpenAI-linked swarms. (sources: Transluce; BleepingComputer)
Transluce — AI Agents Targeted U.S. and Canadian Government Websites (30 Sep 2026)
The Hacker News (1 October 2026) reports on Sucuri research (published 30 September 2026) on a WordPress malware family codenamed SC (from “SC_” markers) described as a blockchain-controlled “self-healing mesh.” Identical backdoor payload is planted in at least eight places — including wp-content/db.php drop-in, a theme-resident twin (e.g. khorshidi/functions.php), fake plugin hyper-engine-kit (and a second copy), database storage, and System V shared-memory segments with a fixed numeric key — so deleting any one copy lets another rewrite the set on next page load or via cron. Code uses a substitution-cipher decoder (no readable function names). Capabilities: hide from plugins screen/update checks; C2 via Ethereum blockchain; fingerprint site; fetch arbitrary JavaScript (skimmers/malware); run PHP; deactivate/delete plugins. Shared-memory copy can survive disk+DB cleanup and on shared hosting may be owned by another account. Initial access vector not confirmed (typical WP plugin/theme flaws, weak creds, supply chain). (sources: The Hacker News; Sucuri) Correction (29 Oct 2026): the Sucuri link pointed to a page that does not exist and now goes to Sucuri's report; internal notes were removed from the title and body. (source: Sucuri) Correction (29 Oct 2026): the card date is now 30 September 2026, the date on Sucuri's linked report, which is the earliest source for this campaign; The Hacker News reported it on 1 October. (source: Sucuri)
The Hacker News — WordPress SC self-healing mesh backdoor / Ethereum C2 (1 Oct 2026)
US Treasury press release sb0640 (30 September 2026; OFAC Recent Actions same day) sanctions Anibal Alexander Canelon Aguirre (a.k.a. “Prometheus” / “The Engineer”) — FBI Ten Most Wanted fugitive alleged to engineer malware used in Tren de Aragua (TdA) ATM jackpotting — plus seven associates and two Mexico-based companies (Enigma Community; Soluciones Integrales Toluca). Scheme: crews break into US ATMs, install malware, remotely force cash dispense; proceeds are laundered, and OFAC's SDN update lists TRON (TRX) digital currency addresses. Treasury: as of August 2025, reported US jackpotting losses US$40.73M across 1,500+ attacks; DOJ has indicted 98 individuals since 21 Oct 2025 with material-support/bank-fraud/computer-intrusion charges (Nebraska and other districts). Designations under E.O. 13581 and E.O. 13224 (TdA is an FTO). (sources: Treasury sb0640; OFAC Recent Actions) Updated 3 Oct 2026 (US DOJ, 2 October US time; 4:34 am 3 October Perth): the Justice Department announced Canelon Aguirre has been apprehended and returned to the US; he made his initial appearance on 2 October before a magistrate judge in the District of Nebraska, entered not guilty pleas and remains detained pending trial. He was indicted there in December 2025 on conspiracy to commit bank fraud, conspiracy to commit money laundering, conspiracy to commit bank burglary and computer fraud, and conspiracy to provide material support to terrorists. DOJ alleges he developed the Ploutus ATM malware, including anti-analysis and self-deletion components, and was a principal leader of the jackpotting conspiracy. He was the first cyber criminal added to the FBI Ten Most Wanted list (March 2026). Updated 5 Oct 2026 (BleepingComputer, 9:01 pm Perth): BleepingComputer cites court documents putting losses above US$5.4 million in at least 63 bank and 54 credit-union jackpottings between February 2024 and December 2025, plus about US$1.43 million attempted. No Australian ATMs are named. Correction (26 Oct 2026): unlinked SecurityWeek and TRM Labs credits have been removed; the TRON addresses come from OFAC's own SDN update, and the OFAC link now points to the 30 September action (it previously pointed to an unrelated 1 October action). The DOJ and BleepingComputer reports now carry Updated lines, and BleepingComputer's report is now linked. (source: US Treasury; OFAC; US DOJ; BleepingComputer)
n8n GitHub security advisories published 30 September 2026 (High; CVSS 4.0 vectors on GHSA pages; no CVE IDs assigned on these GHSAs at publish). Lead issues: GHSA-5jr4-xmvf-frmj — MCP workflow-validation interpreter allows shared prototype mutation so a member with only a read grant can pass permission checks and take over the instance owner account (MFA-protected accounts not affected); GHSA-x8wx-g24x-3549 — Git node Log operation skipped local-config neutralization, enabling program execution as the n8n process user (also usable as an agent tool; also patched on 1.123.83); GHSA-5qpp-pqww-h7fp — Microsoft SQL node v1 interpolates expressions into Query without parameterisation (arbitrary SQL under stored DB credential when untrusted input is bound); GHSA-728h-pmr2-7cgh — Send-and-Wait waiting-webhook HMAC bypass lets a caller with a resume token approve another party’s waiting execution (downstream runs as workflow owner); GHSA-3qcw-p65v-c7vq — unauthenticated unbounded OAuth client persistence via authorize endpoint (disk-fill). Same-day batch also includes additional High/Moderate items (credential check gaps, Chat Trigger XSS, AI Workflow Builder prototype pollution, etc.). (source: n8n GitHub security advisories)
TanStack (Tanner Linsley; blog and GHSA-qx66-fv34-fjm8, 30 September 2026) discloses CVE-2026-102989, a critical reflected cross-site scripting flaw in TanStack Start server-function response handling. The server-function transport can pass request payload fields into internal middleware state instead of limiting them to public input fields; an unauthenticated attacker can craft a server-function URL that returns attacker-controlled HTML from the application's origin. CVSS 3.1 9.3 Critical (AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:N). Affected: published @tanstack/start-server-core, react-start, solid-start, and vue-start from 1.143.12 up to (excluding) patched versions. Fixed: start-server-core 1.169.39; react-start 1.168.60; solid-start 1.168.57; vue-start 1.168.56. Redeploy required after dependency bump. Temporary mitigations: edge/WAF rules on /_serverFn/* (e.g. require x-tsr-serverFn: true; block document navigations); restrictive CSP on server-function responses. (sources: TanStack blog; GitHub Advisory GHSA-qx66-fv34-fjm8)
TanStack Blog — Start security update CVE-2026-102989 (30 Sep 2026)
MetaMask (Consensys) user update (30 September 2026) states the company is responding to an ongoing security incident affecting part of its infrastructure, remediating internally with external partners and advisors. MetaMask says it has identified no immediate threat to MetaMask wallets. As a precaution it is proactively exiting affected validators in its non-custodial staking operations with clients and partners; staking is non-custodial and MetaMask does not manage withdrawal keys for client stake. Lido's research forum disclosure says MetaMask Staking (ex Consensys Staking) began exiting ETH validators in the Lido protocol to reduce network-penalty risk; final validators expected exited (not fully withdrawn) by end of 7 October 2026, with likely foregone rewards and possible downtime penalties. MetaMask has not publicly named the compromised subsystem or confirmed data access. Distinct from wallet-malware phishing campaigns. (sources: MetaMask user update; BleepingComputer)
MetaMask — User update (infrastructure security incident, 30 Sep 2026)
Google DeepMind (Koray Kavukcuoglu; blog 30 September 2026; SecurityWeek 1 October) announces Gemini 4 Argon, a frontier model aimed at long-horizon software engineering, enterprise knowledge work (legal/finance), and cybersecurity defense, with a stated 1M-token context. Initial rollout is to trusted cyber defenders via Google’s Fairwind Program (governments, Google Cloud customers, cybersecurity partners; launched early September with Gemini 3.8 Flash Cyber + CodeMender-style harness). For trusted defenders and Google internal teams, Google says Argon is released without cyber guardrails so they can use full defensive capability; broader developer/enterprise/consumer availability follows further safeguards. Google/Wiz Scan for Good claim Argon found a critical vulnerability exposing sensitive personal information in healthcare software used by hospitals worldwide (software unnamed; remediation status not stated). CWE-bench v1 remediation score cited at 68% (tied first with named peer models per Google). Pricing stated at introductory $2/M input and $10/M output tokens. (sources: Google blog; SecurityWeek)
Google — Gemini 4 Argon: next era of frontier intelligence (30 Sep 2026)
Have I Been Pwned (HIBP) added Medela on 30 September 2026. HIBP says the Swiss medical device company was the target of a ShinyHunters 'pay or leak' extortion campaign in September 2026, and that the data allegedly obtained was later published and included about 424,000 unique email addresses belonging predominantly to healthcare professionals, Medela staff and leads. HIBP says the data was primarily corporate contact information, including names, physical addresses and phone numbers, with some records also containing support tickets. HIBP has flagged the breach as sensitive, so an address can only be checked by its verified owner or through HIBP's domain search. No public statement from Medela was found, and no Australian organisation is named. UPDATE 22 Oct 2026: reworded so each claim is attributed to Have I Been Pwned, and removed wording its breach page does not support (source: Have I Been Pwned). Correction (23 Oct 2026): removed the extortion group's contact deadline, which rested only on the group's own leak post. (source: Have I Been Pwned)
Have I Been Pwned — Medela breach (added 30 Sep 2026; breach September 2026; flagged sensitive)
CVE-2026-61500 (CVE record published 13 July 2026; GHSA-xxrm-3f86-v97j): Rejetto HTTP File Server (HFS) 3.0.0 through 3.2.0 derives the Koa/keygrip session-cookie signing key from non-cryptographic Math.random() and leaks outputs of the same V8 xorshift128+ PRNG to clients during loginSrp1. A remote attacker can collect a small number of login responses, reconstruct PRNG state, recover the signing key, forge an administrator session cookie, and reach RCE via the server_code / administrative API custom-endpoint path. CVSS 3.1 9.8 Critical and CVSS 4.0 9.3 Critical (CVE record). Fixed in HFS 3.2.1 (GitHub release / commit). UPDATE 30 Sep 2026: Horizon3 Attack Research details how Anthropic Mythos (with a custom harness) rediscovered and fully chained the issue — including Z3-based PRNG state recovery and a working exploit path — as an agentic vulnerability-research case study. (sources: GitHub Advisory GHSA-xxrm-3f86-v97j; rejetto/hfs) Updated 5 Oct 2026 (The Hacker News, 4:09 pm Perth): The Hacker News reports that a Python proof-of-concept exploit was publicly released in late September by security researcher Alejandro Ramos (aramosf). Updated 6 Oct 2026 (BleepingComputer, 5 October US time; 4:20 am 6 October Perth): BleepingComputer advises users to upgrade to 3.2.1 or, ideally, the latest stable release, 3.3.4, as soon as possible. No vendor, CISA or other source Cyberstack can cite has confirmed exploitation in the wild; CVE-2026-61500 was not in CISA's Known Exploited Vulnerabilities catalogue when checked on 26 October 2026. No Australian exploitation reports. Correction (26 Oct 2026): reports that exploitation attempts began on 1 October from a China-based actor, that a single China Telecom IP address probed honeypots in Japan and the United States, and that this is the second Rejetto HFS flaw exploited in the wild have been removed: The Hacker News and BleepingComputer attribute them to one threat-intelligence company's honeypot data, and no vendor, CISA or other independent source confirms them. An unlinked SecurityWeek credit has also been removed. The CVSS scores are credited to the CVE record, now linked. (source: CVE record; The Hacker News; BleepingComputer)
