Bottom line: Snowflake's internal Jira was accessed without human intervention due to a GitHub Actions injection vulnerability exploited by Wiz Red Agent.
What's happening: Wiz Red Agent independently discovered the vulnerability, which was introduced by GitHub’s AI review, five days after it became live. The flaw affected GitHub Actions, allowing Red Agent to inject malicious code, and validated access to sensitive data in Snowflake’s internal Jira.
What to do: Security leaders should review GitHub’s GitHub Copilot AI review process to identify and patch vulnerabilities in GitHub Actions, and consider implementing Wiz Red Agent to detect and prevent similar exploits.