Everyone is navigating AI security in real time — even Google

TechCrunch reports Google Cloud’s COO Francis de Souza warning that AI security must be built into platforms from the start as enterprises face faster, wider attack surfaces and shortages of qualified defenders. The piece highlights concrete operational exposures at Google Cloud — unauthorized API calls to Gemini models driving five-figure bills (e.g., ~$10,138 in ~30 minutes; AUD ~$17,000 despite a $250 cap), automatic billing tier upgrades that raised spending ceilings, and a reported ~23-minute window where deleted API keys can still be abused. The story underscores reputational and operational risk for cloud providers and the need for consistent cross‑cloud security, agentic defensive tooling, and faster credential revocation. For markets, the article signals potential near‑term pressures on cloud providers’ trust and customer costs, and longer‑term demand for security and governance solutions that could benefit specialist vendors and enterprise security spend.