Monthly issue · Published October 13, 2026
Fifteen picks across five lanes. A wormable DNS bug to finish patching, ransomware crews hitting healthcare and government, a new Wazuh on the way, incident-reporting clocks getting closer, and AI agents leaking data in new ways.
September's Patch Tuesday fixed CVE-2026-69730, a Windows DNS remote-code-execution flaw described by ZDI as a successor to SigRed, plus two exploited privilege-escalation bugs: CVE-2026-81963 (Windows Update Stack) and CVE-2026-85880 (ALPC).
Microsoft's October release lands on the 13th. Check whether any fixes are flagged as exploited or publicly disclosed before deciding your rollout order.
CISA added CVE-2026-55040 (SharePoint weak authentication), CVE-2026-59310 (vCenter path traversal), CVE-2026-33824 (Windows IKE double free) and CVE-2026-65400 (macOS improper authentication) to the Known Exploited Vulnerabilities catalog.
Tracked incidents over the summer were led by healthcare/medtech and public-sector victims. Active names include ShinyHunters, Qilin, Clop and Rhysida, with high-profile hits on Boston Scientific, Manchester Airports Group and the Berlin city government.
In a reported capture-the-flag evaluation, an AI model guessed passwords and reused credentials found in public repositories to reach real company systems. Leaked secrets are now cheaper than ever to exploit.
A reported plugin-installation flaw affecting several AI coding tools let pinned plugins be swapped for malicious code without prompting the user.
Wazuh 5.0 betas add a built-in threat-intel (CTI) section, incident response in the dashboard, AI assistant settings, and switch file-integrity who-data to eBPF by default. Manager daemons and config files were reorganized, and legacy OS support was dropped.
With a wormable DNS bug patched in September, unpatched DNS servers are an obvious target. Exploitation of server processes usually shows up as crashes and unusual child processes.
A reported flaw in the official MCP Python SDK let a malicious MCP server capture OAuth credentials. Agent tool calls are becoming a new privileged-access path.
CISA's final CIRCIA rule was slated for September 2026 in the Unified Agenda. When it takes effect, covered critical-infrastructure entities must report covered incidents within 72 hours and ransom payments within 24 hours. Check the Federal Register for publication status and the effective date.
Under the Digital Omnibus deal, high-risk obligations for stand-alone (Annex III) systems move to December 2, 2027, and for AI inside regulated products (Annex I) to August 2, 2028. The Parliament approved it in June; confirm final Council adoption and Official Journal publication.
Exploited vulnerabilities on the CISA KEV list are a clean piece of audit evidence: they show whether your patch-management control actually works on deadline.
Reports at the end of September described AI coding agents publishing more than 13,000 internal screenshots to public GitHub repositories, exposing billing and treasury data from hundreds of organizations.
A flaw in the official MCP Python SDK let a malicious MCP server capture OAuth client credentials and authorization codes.
Microsoft patched Reprompt, a Copilot issue where one click on a legitimate link could silently exfiltrate data, and CVE-2026-85889, a missing-authentication flaw in Azure AI Foundry that Microsoft mitigated with no customer action required.
Some items come from secondary trackers and news reports. Confirm CVE details and dates with the vendor or agency advisory before acting.
Free · once a month
One email a month with the curated picks for Blue, Red, Purple, Compliance and AI, each with a "do this now" step.