Google has expanded its Cloud Vulnerability Reward Program (Cloud VRP) by launching a new top reward tier of up to $250,000, dedicated to architectural vulnerabilities that could lead to comprehensive administrative control over its cloud environments and bypass customer isolation.
Conditions for obtaining the major reward
Google explained that the reward targets advanced architectural security research that demonstrates the ability to seize cloud project and organization administration privileges ('roles/owner') or impose full Identity and Access Management (IAM) control across any organization or project within Google Cloud without prior access conditions. To qualify for this reward, the company requires that the exploitation path does not rely on stolen accounts, prior access, social engineering, or human interaction methods.
Evaluation criteria also include vulnerabilities that allow remote code execution (RCE) in the Borg system, which is the core production infrastructure for Google's ecosystem, or obtaining root-in-production privileges.
Strict rules of engagement to prevent service disruption
Given that unauthorized code execution within the Borg system threatens service stability and the data security of multi-tenant subscribers, Chris Betz, Chief Information Security Secretary at Google Cloud, emphasized the necessity of completely stopping tests once initial code execution is proven. He stressed that any attempts to escalate privileges, extract production credentials, or perform lateral movement within the network will immediately lead to the disqualification of the researcher and the cancellation of the reward.
To protect actual customer accounts from operational risks during penetration testing, Michael Couts, head of the cloud vulnerability reward program at Google, pointed out the design of an inevitable verification mechanism known as Proof-of-root, which relies on a dedicated service account, allowing researchers to provide cryptographic proof of their unauthorized access against that identity without compromising user data.
The impact of artificial intelligence in security research
Couts explained that discovering such architectural flaws and bypassing tenant isolation today combines human intuition with artificial intelligence-based tools and agentic systems. Google's data showed that new researchers who integrate their personal intuition with agentic systems are now able to discover more complex vulnerabilities compared to previous years, prompting the company to raise the value of financial incentives to keep pace with this development in scanning and structural analysis capabilities.