NewsIncident
Microsoft: ARTEX agents search GitHub for leaked keys, then use them in Azure; in one case secret search to Key Vault theft took five minutes
Microsoft Threat Intelligence and Microsoft Security Research report that since August 2026 threat actors have used ARTEX, an agentic AI offensive framework, to send large volumes of GitHub searches that combine a target organization's name with words for secrets, cloud services or AI providers. The agents read the results and ran follow-on attacks with little hands-on work. Secrets found in public repositories included OAuth application client secrets, application credentials and AI inference API keys, left in configuration files, scripts or code; the agents also searched Azure DevOps. From October 28 sign-ins with service principals reached dozens of applications in several tenants, some through a Tor exit node, going after storage account keys, App Service settings, connection strings, AI connections and AKS credentials. In one case an agent went from a GitHub search to taking secrets from Key Vault in five minutes. Microsoft does not name the actors. It advises rotating any exposed credential at once, revoking sessions and tokens, turning on GitHub secret scanning and push protection, and moving OAuth apps to managed identities.
Why it matters
A key pushed to a public repository can now be found and used within minutes. Turn on secret scanning and push protection, and treat any leaked key as already used.