Google launches Scan for Good: AI agents hunt critical infrastructure flaws

Google unveiled Scan for Good, using Gemini 3.8 Flash Cyber and Wiz's Red Agent to autonomously find security flaws in hospitals, a municipality, a rail operator, and major tech providers. All findings are validated by human researchers before disclosure.

• Initiative pairs Gemini 3.8 Flash Cyber with Wiz's Red Agent
• AI found a GitHub Actions flaw in a Snowflake repo, fixed same day
• Findings include exposure of 8.8M files and 5,000 residents' data
• Humans validate every finding; program scales globally