CISOS Debate Human Role in AI-Powered Security covers a RSAC 2026 Conference panel where Google Cloud, Vodafone and PayPal discussed evolving AI use cases and how to deploy them safely. The panel, featuring Francis deSouza, Emma Smith and Shaun Khalfan, examined whether AI deployments need a “human in the loop” or if people merely slow progress.
One surprising note was that Google reportedly has 50% of its code AI generated with developer assistance, while Vodafone uses AI to automate workflows and even generate board summaries; PayPal uses AI to help detect fraud across billions of transactions each month. The discussion also covered the idea of placing a human “on the loop” to gain insights from AI rather than trying to control every step, with Smith arguing that a human in the loop is not scalable for long‑term, scaled operations.
DeSouza warned that human‑led defenses can be too slow to stop agent‑led cyberattacks, signaling a shift toward agent‑led defence. Khalfan emphasised data security and governance, noting PayPal ranks AI models by data sensitivity and the need for a risk and compliance wrapper around deployments. According to The Wall Street Journal, the conference panel explored how leadership and collaboration, including industry initiatives like CoSAI, shape secure, ethical AI adoption.