Google is grappling with security challenges as AI integration accelerates, highlighted by recent issues with its Gemini API billing. Google Cloud COO Francis de Souza emphasized the need for simultaneous advancement of AI, data, and security strategies, warning against "shadow AI" where employees bypass governance using consumer AI tools. This complicates unified auditing and data management, with the average time from breach to attack progression now just 22 seconds. Meanwhile, Google Cloud developers faced unexpected bills due to unauthorized Gemini API calls, linked to publicly exposed API keys. These keys, initially for Google Maps, were exploited after Google expanded their access scope without clear developer notification. Despite issuing refunds, Google maintains its policy on automatic billing tier upgrades, prioritizing service continuity over budget limits. Security firm Aikido noted a 23-minute vulnerability window post-key deletion, though newer credential formats show improved revocation times.