Alberta Government Found Vulnerabilities in 466 Million Lines of Code Using Claude

Anthropic published a case study on July 6, 2026 of the Canadian province of Alberta's government, which since 2025 has used Claude Code with Claude Opus and Claude Sonnet models to find and fix vulnerabilities in its state IT systems.

What Alberta's team actually did

A team within Alberta's Ministry of Technology and Innovation deployed Claude Code to scan the codebase maintained by the ministry for all 27 provincial ministries—ranging from social services to public safety and wildfire response. Approximately 50 agents worked autonomously and in parallel, checking systems for security vulnerabilities, infrastructure weaknesses and deployment process gaps, as well as gaps in technical documentation.

• Publication date of Anthropic's case study — July 6, 2026
• Volume of code reviewed — 466 million lines in 20 hours
• Number of autonomous Claude Code agents — approximately 50, working in parallel
• Scale of the ministry's systems — 27 provincial ministries, approximately 1,280 applications and 3,400 code repositories
• Models — Claude Code based on Claude Opus and Claude Sonnet

Why Alberta's government needed this

Much of the ministry's codebase had never undergone systematic security review, and the accumulated technical debt—unsafe code, unfixed bugs, outdated software—is estimated at billions of dollars. At the same time, the ministry's systems store extremely sensitive data: tax records, government procurement data, and social services case files.

"Albertans trust their government with some of the most sensitive data in their lives, and our obligation is to protect it.

By using AI to find and fix vulnerabilities across all our systems, we accomplished in hours what would have taken years using traditional approaches. This is what responsible governance looks like in the age of AI, and the best is yet to come," said Nate Glubish, Alberta's Minister of Technology and Innovation.

Alberta also published a series of technical white papers describing its experience so that other governments could benefit from it.

What it means

Alberta's case is one of the first public examples of a regional government using AI agents for systematic security audits of hundreds of millions of lines of inherited state code—not for isolated pilot projects—and demonstrates that such audits are now measured in hours, not years.

Frequently Asked Questions

How much code did

Alberta's team review and in what timeframe?

Approximately 50 agents based on Claude Code with Opus and Sonnet models scanned 466 million lines of code in 20 hours, covering approximately 1,280 applications and 3,400 repositories across 27 provincial ministries.

Since what year has Alberta used Claude Code for security?

The team at Alberta's Ministry of Technology and Innovation has been applying Claude Code since 2025 to find and fix vulnerabilities, infrastructure weaknesses, and documentation gaps in state systems.