GitLab shipped agentic AI security remediation reaching general availability

GitLab released Agentic SAST Vulnerability Resolution reaching general availability, allowing AI agents to analyze SAST findings, generate code fixes, and create ready-to-merge requests automatically. The platform introduced new agents for CI pipeline setup and delivery analytics, expanding AI across the DevSecOps lifecycle with credit-based cost controls.

Microsoft unveiled an agent-first platform at Build 2026 with new GitHub Copilot App, Project Solara security framework, and MDASH multi-model agentic security system. The update emphasizes autonomous AI agents for DevOps workflows with enhanced security controls and enterprise governance.

GitProtect.io report reveals 236 vulnerabilities patched across major DevOps platforms in 2025, with 59% rated high or critical severity. Platforms including GitHub, GitLab, Azure DevOps, and Atlassian saw accelerating vulnerability discovery, particularly in Q4 with 76% increase over Q1.

Atlassian extended data export rules to block file downloads from Jira and Confluence attachments, enhanced FedRAMP environment self-service capabilities, and introduced Connect framework end-of-support warnings. These changes focus on preventing data leakage and improving enterprise compliance controls.

Microsoft resolved service-side problems affecting Bitbucket integration with Azure DevOps App Service after API deprecation issues on April 8. Organizations experienced connection failures with Bitbucket Cloud service connections, requiring updated authentication methods.

GitHub launched Medium-tier code reviews in Copilot, routing complex pull requests to higher-reasoning models for deeper security analysis. The update includes sandboxed execution environments for agentic development, improved VS Code agent windows, and GPT-4.1 deprecation in favor of GPT-5.5.