The Information Machine

Government of Alberta uses Claude to find and fix cybersecurity vulnerabilities across government systems

Anthropic News · 2026-07-06

The Government of Alberta used Claude Code with Opus and Sonnet models to scan 466 million lines of code across 27 provincial ministries in 20 hours, identifying and remediating cybersecurity vulnerabilities that traditional methods would have taken an estimated 6.5 years to review.

Open original ↗

Appears in

Extraction

Topics: ai-for-governmentcybersecurityclaude-codeai-agentslegacy-modernization

Claims

  • Alberta's Ministry of Technology and Innovation deployed approximately 50 parallel Claude Code agents to scan 466 million lines of government code in 20 hours, a task estimated to take 6.5 years through traditional approaches.
  • Claude Code not only identified vulnerabilities but also generated patches, wrote missing automated tests, and in some cases rebuilt legacy systems in modern languages within four to five days.
  • Alberta built continuous red-team and blue-team Claude agents on the Claude Agent SDK that check each application against approximately 95 security controls per pass.
  • Alberta published technical white papers documenting its approach so other governments can replicate the effort, and is hosting an industry day in Edmonton in July 2026.
  • Alberta plans to use Claude Code to consolidate 185 legacy applications in one ministry down to 16 reusable modern applications.

Key quotes

By using AI to find and fix vulnerabilities across our systems, we accomplished in hours what would have taken a traditional approach years to complete. This is what responsible government looks like in the AI era.
Around 50 agents worked autonomously and in parallel to scan the systems for security vulnerabilities, weaknesses in underlying infrastructure and deployment processes, and gaps in technical documentation.
In some scenarios, these systems could be rebuilt in as little as four to five days, including a subsidy program portal that was originally hand-coded in Java roughly 25 years ago and took five months to build the first time.