More talk and Perspective about Project Perception or is that “Inception” (see what we did there) from Mona Ghadiri - a Microsoft MVP, Capgemini Distinguished Cyber Engineer | Zero To Hero Board Member | MSFarsi Leader | MGCI Regional Lead
We will ask her to give us her “highly opinioned” facts on the OpenAI Agent “mishap” involving Hugging Face and the evolving world of Cyber-AI-Security.
Words of Wisdom:
“You will be judged on how well you treat those who can do nothing for you.”
Security Insights - Foresight - Hindsight – August 2026 Edition
General
Lots of free tech training - Explore events at Microsoft
Rethinking security for the age of AI – Introducing Project Perception | Microsoft Blog
What’s new in Microsoft Security: July 2026 | Microsoft Security Blog
Securing our future: July 2026 Secure Future Initiative progress report | Microsoft Security Blog
AI Security
OpenAI autonomous agent escape and Hugging Face incident | Hugging Face + OpenAI disclosure
Enhancing AI security through global AI red teaming | Microsoft Security Blog
Agent365 / Agentic Security – Project Perception Deep Dive
What is Project Perception? | Microsoft Learn
Get started with Project Perception | Microsoft Learn
Project Perception product page | Microsoft Security
Rethinking security for the age of AI | Microsoft Blog
Introducing MAI-Cyber-1-Flash inside MDASH | Microsoft AI
Key Project Perception Details
Project Perception is Microsoft’s new multi-agent security system that coordinates specialized AI agents across three roles:
Red team agents – continuously map attack paths and probe for weaknesses before adversaries can exploit them
Blue team agents – investigate signals, correlate context, and prioritize real risk
Green team agents – remediate findings and harden the environment
It runs as a closed-loop system that reasons over Microsoft security signals, organizational context, and threat intelligence. High-impact actions remain under human control. Public preview began August 3, 2026, initially inside the Microsoft Defender portal, with plans to expand across the Microsoft Security portfolio. Pricing uses Security Compute Units (SCUs).
Azure Security & Defender for Cloud News
What’s new in Defender for Cloud | Microsoft Learn
Threat Intelligence
CaptiveCrunch: Midnight Blizzard targeting travelers | Microsoft Security Blog
Email threat landscape: Q2 2026 trends | Microsoft Security Blog
Microsoft Entra
Microsoft Entra ID security updates – Passkeys as default | Microsoft Security Blog
Device Management & Protection (Intune)
What’s new in Microsoft Intune | Microsoft Learn
Defender XDR & Sentinel
Monthly news – July/August 2026 | Microsoft Defender XDR Blog
Sentinel Graph tools + custom detection as code | Microsoft Learn
Defender XDR + Sentinel unified operations | Microsoft Learn
Copilot for Security
Security Copilot agentic capabilities | Microsoft Learn
Purview – Compliance & Governance
Purview data protection for AI agents | Microsoft Learn
Purview for Agent 365 | Microsoft Learn
Non Microsoft Security News
OpenAI agent sandbox escape that compromised Hugging Face and additional third-party accounts → Hugging Face report + OpenAI statement
AI for the Masses
LiteLLM and open-weight model risks
Model ablation / safety-rail bypass techniques
Embedding space attacks
Agent pentesting and bug-bounty trends
Featured Resources & Deep Dives
What’s New in Defender (August 2026)
What’s new in Microsoft Defender XDR | Microsoft Learn
Project Perception public preview (Red / Blue / Green agent teams for attack simulation, investigation & remediation) – available in Microsoft Defender starting August 3
MAI-Cyber-1-Flash integrated with MDASH – specialized cybersecurity model delivering ~96% on CyberGym at roughly half the previous cost by handling ~90% of routine tasks
Daily Defender Dispatch – August 13, 2026
Daily Defender Dispatch: Project Perception Public Preview Live + MAI-Cyber-1-Flash
1. Project Perception Public Preview is Live
Microsoft’s new multi-agent security system entered public preview on August 3 inside the Microsoft Defender portal.
It coordinates three specialized agent teams:
Red – continuous attack-path mapping and proactive probing
Blue – investigation, prioritization, and detection engineering
Green – remediation and hardening
Humans retain final control over high-impact actions. Access it via the Perception blade in the Defender portal.
→ Announcement | Learn overview | Get started
2. MAI-Cyber-1-Flash + MDASH
Microsoft’s first in-house cybersecurity model (MAI-Cyber-1-Flash) is now powering the MDASH multi-agent vulnerability harness. It handles the majority of routine tasks and escalates only the hardest work to larger models (e.g., GPT-5.4), delivering top-tier CyberGym performance at significantly lower cost.
→ MAI-Cyber-1-Flash announcement
3. Why This Matters
Project Perception represents Microsoft’s clearest move yet from “AI that assists” (Security Copilot) to “AI that acts” under human oversight. Early adopters should evaluate it for vulnerability management playbooks first, then expand to threat-intel driven investigations.
Takeaway for defenders:
If you already have Defender XDR, log into the portal today and explore the new Perception experience. Start with low-risk playbooks and keep human approval gates enabled.
