Back to updates

Security - 6

2026 (12)

March (10)

security: Observability for AI Systems: Strengthening visibility for proactive risk detection

Microsoft Security Blog

2026-03-18

The post says observability is becoming necessary as AI systems become more autonomous, because visibility into AI behavior can help detect risk earlier and support secure development. It is presented as a Microsoft Security Blog article focused on proactive risk detection.

github: Dependabot now detects malware in npm dependencies

GitHub

2026-03-17

Dependabot can now alert on npm dependencies that match known malware advisories. When malware alerting is enabled, it checks repository npm dependencies against malicious package versions.

github: Secret scanning in AI coding agents via the GitHub MCP Server

GitHub

2026-03-17

GitHub MCP Server can now scan code changes for exposed secrets before commit or pull request creation. The feature is intended to catch credential leaks early in AI coding agent workflows.

security: New Microsoft Purview innovations for Fabric to safely accelerate your AI transformation

Microsoft Security Blog

2026-03-16

Microsoft announced new Microsoft Purview innovations for Fabric aimed at supporting safer AI adoption by strengthening security and governance. The post frames Purview as part of the governance layer for organizations using AI with Fabric.

security: Contagious Interview: Malware delivered through fake developer job interviews

Microsoft Security Blog

2026-03-11

Microsoft describes the Contagious Interview campaign, where attackers impersonate recruiters from crypto and AI firms to target developers with fake coding assessments. The delivered malware installs backdoors and steals API tokens, cloud credentials, crypto wallets, and source code.

k8s: Announcing the AI Gateway Working Group

Kubernetes Official Blog

2026-03-09

Kubernetes has formed a new AI Gateway Working Group to define standards and best practices for Gateway API-based networking infrastructure tailored to AI workloads.

security: Secure agentic AI for your Frontier Transformation

Microsoft Security Blog

2026-03-09

Microsoft Security Blog post about securing agentic AI initiatives, highlighting Microsoft Agent 365 and Microsoft 365 E7 in the context of a “Frontier Transformation.”

security: Malicious AI Assistant Extensions Harvest LLM Chat Histories

Microsoft Security Blog

2026-03-05

Microsoft reports a campaign of malicious AI browser extensions that exfiltrated LLM chat histories and browsing data from services including ChatGPT and DeepSeek, affecting large numbers of users and enterprises.

github: Dependabot alert assignees are now generally available

GitHub

2026-03-03

GitHub Dependabot now supports assigning vulnerability alerts to specific users (generally available). This adds per-alert ownership to help teams track and remediate dependency vulnerabilities.

security: OAuth redirection abuse enables phishing and malware delivery

Microsoft Security Blog

2026-03-02

Microsoft reports attackers abusing OAuth redirect behavior to move users from legitimate sign-in pages to attacker-controlled sites, enabling phishing and malware delivery via trusted authentication flows.

February (2)

azure: [In preview] Public Preview: Restrict usage of user delegation SAS to an Entra ID identity

Azure Updates

2026-02-26

Azure Storage adds a preview feature to restrict (bind) user delegation SAS usage to a specific Microsoft Entra ID identity, strengthening authentication by tying SAS access to the user.

security: Threat modeling AI applications

Microsoft Security Blog

2026-02-26

Microsoft Security Blog notes that AI threat modeling is used to identify misuse, emergent risks, and failure modes in probabilistic and agentic AI systems.