Threat actors use agentic AI tool ARTEX to hunt exposed secrets and abuse cloud access
- Severity
- High
- Sources
- 1
What happened
Microsoft observed the agentic offensive framework ARTEX scanning GitHub from August 2026 for cloud, AI inference and application secrets, then chaining follow-on actions including Azure Key Vault and Storage Account exfiltration and Azure AI services abuse. A compromised vendor-owned multitenant application credential was used against downstream tenants, and Microsoft Entra ID sign-ins came from 199.19.77.84 and 199.19.77.18, the latter a Tor exit node. Defenders should assume exposed secrets are compromised and rotate them, as removing the commit does not remediate exposure; they should also enable GitHub public monitoring and push protection and move OAuth applications to managed identities. No CVE is identified.
For security analysts · Written by AI from 1 source · Wrong?
What to do
- Assume any secrets found in public repositories are compromised.
- Rotate exposed keys and credentials immediately and remove them from the codebase.
- Scan GitHub repositories for secrets and enable public monitoring and push protection.
- Move OAuth applications away from secret-based authentication to managed identities.
Written by AI from the sources, checked against them
Full story timeline · 1 update
Who reported · 1 outlet
Who and what is in it
Companies