Find articles
20 results for “azure”
How Storm-3068 turned a leaked credential into full Azure tenant takeover
Attackers used a compromised developer identity to pivot from source code repos to cloud admin rights, exposing gaps in identity hygiene and pipeline security.
Google Cloud Next ‘26: AI agents, Wiz integration, and new security controls
Google Cloud unveiled AI-powered security agents, Wiz’s expanded AI protection, and new Trusted Cloud capabilities at Next ‘26.
Azure Developer CLI Extension Framework GA Enables Custom Dev Workflows
The Azure Developer CLI (azd) extension framework is now generally available, allowing teams to build and share custom workflows for project initialization, deployment, and validation.
Microsoft DCU disrupts EvilTokens PhaaS platform behind device code phishing surge
Microsoft’s Digital Crimes Unit disrupted EvilTokens infrastructure, a top PhaaS platform using AI-assisted lures and automated token theft for device code phishing attacks.
Google Cloud Next '26 launches cross-cloud infrastructure for agentic AI workloads
Google Cloud unveiled new capabilities across compute, networking, data, and sovereignty to support AI agents at machine speed, including Agent Gateway, Smart Storage, and Confidential External Key Management.
Azure MCP Server 2.0 GA Delivers Self-Hosted Agentic Automation
Azure MCP Server 2.0 reaches general availability with self-hosting support, security hardening, and performance gains for agentic workflows.
Microsoft releases Cloud Web Applications Threat Matrix aligned with MITRE ATT&CK
Microsoft introduces a MITRE ATT&CK-aligned threat matrix to map and mitigate risks in cloud-hosted web apps and serverless platforms.
Microsoft Defender adds AI-themed attack detection for phishing and malware
Microsoft Defender now identifies and disrupts AI-generated phishing, malware, and multi-stage attacks across the kill chain using behavioral and content analysis.
ASCII smuggling shifts from AI prompt injection to phishing evasion
Attackers now use invisible Unicode characters to bypass email filters, repurposing a technique once used to hide prompts from AI models.