Google Cloud has introduced a $250,000 “Root-in-Cloud” reward for security research that demonstrates full administrative control of arbitrary Google Cloud organizations or projects, the company said October 29. The new top-tier payout is aimed at architectural weaknesses such as tenant-isolation bypasses and complex cross-service logic flaws rather than routine bugs.
Researchers must show control equivalent to the roles/owner role or complete IAM authority, starting without access to the target project or any special project configuration. Google’s eligibility rules exclude attack paths dependent on leaked, phished or stolen credentials, as well as user interaction and social engineering. Google Cloud’s announcement describes the reward as an expansion of its existing Vulnerability Reward Program. [2]
The company is also extending consideration for the reward to valid reports involving remote code execution in Borg, the production infrastructure underlying Google services, or “root-in-prod” access. Google said the Cloud VRP panel will assess the initial entry point, execution context and attack surface to determine whether the finding could enable arbitrary tenant or project takeover.
Google is imposing strict limits on testing because a Borg compromise could affect service availability or expose data across customer boundaries. Researchers must stop probing immediately after achieving code execution in Borg; escalation, production-credential extraction and lateral movement are prohibited and can disqualify a submission. The company said it has created a designated service account and a deterministic, cryptographic proof-of-root process so researchers can validate unauthorized access without testing against real customer projects.
The announcement is not tied to a disclosed CVE, patch or known exploitation campaign. It instead reflects Google’s effort to encourage deeper, human-led research as automated tools increase the volume of vulnerability reports. Defenders should not interpret the bounty as evidence of an active Google Cloud flaw, but organizations should continue enforcing least-privilege IAM, monitoring unusual privilege changes and reporting suspected weaknesses through the Google Cloud VRP rules rather than attempting unapproved testing.
