Wiz says its security agent used a public software workflow to reach Snowflake’s internal Jira records.
Source: WizWorkflow exposed June 18; research June 23; publication August 17.
Editorial analogy: a public suggestion box accidentally operates a staff-only filing cabinet. The lesson concerns a boundary between public input and internal privileges; this was a digital workflow.
An analogy for the mechanism; not an additional claim about the incident.An exposed credential allowed reading engineering, compliance and bug-bounty records. Wiz
AI authorship of the vulnerable code is unconfirmed.
Snowflake reported no evidence of access beyond this research.
Editorial lesson: automated checks need independent review, and a single workflow should hold only the permissions it needs.
Researchers reported through Snowflake’s disclosure program and deleted accessed data. Wiz
Workflow patched June 23; token rotated June 24, according to the detailed timeline. Wiz
Editorial defensive connections. These describe useful control categories, not evidence that a named product would have prevented this research finding.
Treat public input as untrusted data. Review how automated jobs process it, and test important security boundaries independently of routine code checks.
Automated analysis can miss context. Testing needs appropriate authorization and safe stopping rules.
Give each automated job only the permissions it needs. Assign owners to credentials, review their reach and prepare to revoke exposed keys across connected services.
Replacing a key does not repair the path that exposed it. Access review and investigation remain separate tasks.
Editorial connections to relevant controls, not evidence that a particular technology would have prevented this case. Each guide links to the security guidance behind its recommendations.
Explore the full AI security frameworkExplore more original accounts in the source report library ↗.
These are source-reported findings. An independent assessment, when available, is labeled explicitly.
01Read this case’s update history →
Reviewed Sep 29, 2026 · Editorial methodology · Structured data