Threat modeling AI applications
AI threat modeling helps teams identify misuse, emergent risk, and failure modes in probabilistic and agentic AI systems.
Research
Explore in-depth research on the latest cybersecurity threats, trends, and defense strategies. Get insights from Microsoft that’ll help you better understand and respond to today’s challenges.
Refine results
Topic
Products and services
Publish date
-
-
Developer-targeting campaign using malicious Next.js repositories
A developer-targeting campaign leveraged malicious Next. -
Running OpenClaw safely: identity, isolation, and runtime risk
Self-hosted agents execute code with durable credentials and process untrusted input. -
Detecting and mitigating common agent misconfigurations
Agents are increasingly powerful. With that power comes risk: small misconfigurations, over‑broad sharing, unauthenticated access, and weak orchestration controls can create real exposure. Retain Microsoft Security Experts
Microsoft Security Experts are now available to strengthen your team with managed security services. Learn how to defend against threats with security experts.
-
Manipulating AI memory for profit: The rise of AI Recommendation Poisoning
That helpful “Summarize with AI” button? It might be secretly manipulating what your AI recommends. -
A one-prompt attack that breaks LLM safety alignment
As LLMs and diffusion models power more applications, their safety alignment becomes critical. -
Analysis of active exploitation of SolarWinds Web Help Desk
We are seeing exploitation of SolarWinds Web Help Desk via CVE‑2025‑40551 and CVE‑2025‑40536 that can lead to domain compromise; here is how to patch, hunt, and mitigate now. -
New Clickfix variant ‘CrashFix’ deploying Python Remote Access Trojan
CrashFix crashes browsers to coerce users into executing commands that deploy a Python RAT, abusing finger. Modernize your Security Operations Center with Microsoft Sentinel
Microsoft Sentinel is a cloud-native SIEM solution powered by AI and automation that delivers intelligent security analytics across your entire enterprise.
-
Detecting backdoored language models at scale
We’re releasing new research on detecting backdoors in open-weight language models and highlighting a practical scanner designed to detect backdoored models at scale and improve overall trust in AI systems. -
Infostealers without borders: macOS, Python stealers, and platform abuse
How modern infostealers target macOS systems, leverage Python‑based stealers, and abuse trusted platforms and utilities to distribute credential‑stealing payloads. -
Case study: Securing AI application supply chains
Securing AI-powered applications requires more than just safeguarding prompts. -
Turning threat reports into detection insights with AI
Security teams often spend days manually turning long incident reports and threat writeups into actionable detections by extracting TTPs.
