The AI & Security Insights Show
The AI & Security Insights Show
The AI & Security Insights Show·Nov 20, 2026·1h 35m

The AI & Security Insights Show - Episode 300!!! - Cyclotron | Agent365 Proof-cases.

Show notes

Security Insights – Foresight – HindsightNovember 19, 2026 Edition

This is our 300th Episode!!! We celebrated it with our friends from Cyclotron. Nathan returns as guest to discuss their Agent 365 PoC offering with lessons learned from a structured approach to proofing out Agent 365 capabilities.

Security Insights – Foresight – HindsightNovember 19, 2026 Edition

The crew continues in the regular cadence. Focus remains high-signal Microsoft security coverage: agent risks, collapsing patch windows, and practical controls.

General / Microsoft Security

* What’s new in Microsoft Security: November 2026 (Nov 5) — Dynamic Permissions, Microsoft Execution Containers (MXC) GA, agent runtime protection, local AI agent discovery expansion, Exposure Management updates.

* Microsoft named a Leader in the 2026 Gartner Magic Quadrant for AI Agent Management Platforms (Nov 3)

* November Patch Tuesday: ~1,094–1,259 CVEs fixed (second consecutive month over 1,000), including an exploited Outlook flaw that can leak NTLMv2 hashes (CVE-2026-124194).

* Insights from the 2026 Microsoft Digital Defense Report

AI Security

* The risks of device code authentication for AI coding agents (Nov 17)

* Trained to betray: How post-training supply chain attacks hijack coding agents (Nov 12) — poisoned datasets and checkpoints leading to RCE and other actions.

* Lessons from 59 Linux Kernel Bug Reports (FORGE Lab / MDASH) (Nov 11)

Agent365 / Agentic

* Dynamic Permissions and least-privilege boundaries via MXC (GA).

* Local AI agent and MCP discovery expansion on Windows and macOS (GA).

* Agent runtime protection settings generally available.

* Continued Project Perception / Guardian Agent and Dynamic Agent Risk work.

Threat Intelligence

* High volume phishing campaign uses Microsoft 365 outage lures to target multiple industry sectors (Nov 16)

* SCKit abuses ScreenConnect to deploy remote access and target technician sessions (Nov 13)

* The global socket: how threat actors are weaponizing GSocket for stealthy C2 (Nov 11)

* Threat actors use agentic AI tooling to find and abuse exposed secrets and ARTEX abuse expands as cybercriminals target financial sector (Nov 5–6)

Microsoft Entra

Identity remains the control plane. Device-code and related authentication risks for agents highlighted; continue enforcing phishing-resistant MFA, Conditional Access, and least privilege for both humans and agents.

Defender XDR & Sentinel

ISOC in the Defender portal continues in preview (eligible E5/E7/Defender Suite tenants without an active Sentinel workspace). Starting November 15, existing Sentinel customers meeting licensing criteria can opt in; included retention for eligible Defender data moves toward 90 days. Non-Microsoft data still requires a workspace and is billed separately.Additional notes: expanded local AI agent discovery, synthetic media detection events in advanced hunting, and ongoing agent posture / runtime protections.

Purview – Compliance & Governance

Sensitivity labels, DLP, and auditing continue extending to agent interactions and data access.

Non-Microsoft Security News

Continued coverage of PHP2Shell aftermath, agent-driven incidents, large crypto heists attributed to North Korean actors, and supply-chain events involving coding agents and RMM tools such as ScreenConnect.

AI for the Masses

Practical agent hygiene, open-weight and coding-agent risk management, and the reality that detection, containment, and least privilege matter more than perfect patching speed amid record CVE volumes.

Featured Resources

* 2026 Microsoft Digital Defense Report

* What’s new in Microsoft Security: November 2026

* Project Perception / Guardian Agents and Dynamic Permissions documentation

* Rod Trent Substack and related daily check-ins

What’s New in Defender

Dynamic Permissions, MXC GA, local AI agent + MCP discovery (Windows/macOS), agent runtime protection GA, ISOC preview progress, synthetic media detection events, and expanded identity/AD CS-related detections.

* Full archive and live links: www.microsoftsecurityinsights.com



This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.microsoftsecurityinsights.com