Google has introduced a $250,000 reward in its Cloud Vulnerability Reward Program (VRP) for severe architectural security flaws, including tenant-isolation bypasses and complex cross-service bugs that could give an attacker full administrative control over Google Cloud organizations or projects.
To qualify, researchers must demonstrate control equivalent to the “roles/owner” permission or complete IAM control over arbitrary organizations or projects, without prior access or special project setup. The exploit must not rely on compromised credentials, social engineering or user interaction.
Google Cloud CISO Chris Betz said reports showing remote code execution in the company’s underlying production infrastructure, known as Borg, or achieving root-in-prod will also be assessed for eligibility. Researchers who gain execution on production infrastructure must stop testing: further escalation, attempts to obtain production credentials or lateral movement are prohibited and can lead to disqualification.
Google says researchers should not test against real customer projects. Instead, it has introduced a deterministic proof-of-root process that uses a designated service account, allowing researchers to cryptographically demonstrate unauthorized access. Program rules provide endpoint specifications, JSON payloads and command-line examples.
The expanded reward follows Google’s decision to suspend submissions to its Open Source Software Vulnerability Rewards Program in October after receiving a flood of AI-generated reports. Since launching its first vulnerability reward program in 2010, Google says it has paid researchers more than $81.6 million. In 2025, the company awarded $17.1 million to more than 700 researchers, up from $12 million in 2024.
Sources: BleepingComputer.