Centre stage
Panel

Delegated decisions, amplified risks: Charting a secure future for agentic AI

In person & online
  • Date
    8 July 2025
    Timeframe
    11:05 - 11:35 CEST
    Duration
    30 minutes
    Share this session

    As “magic genie bots” that promise to manage our daily lives, AI agents are poised for rapid deployment. But what is the true cost of this convenience? Join Meredith Whittaker and Kenneth Cukier at the ITU AI for Good Summit 2025 for a critical conversation on the profound risks and emergent solutions in agentic AI.

    This session will move beyond the hype to examine the core risks of AI agents: the unprecedented “root” access they require to our most sensitive personal data. The discussion will explore how granting autonomous systems access to our calendars, messages, and financial information creates a honeypot for misuse and surveillance, effectively asking us to hand over our digital brains to third parties.

    Whittaker and Cukier will debate the potential for catastrophic privacy erosion, the illusion of user control, and the blurring of boundaries between applications and operating systems. They will also outline a path forward, debating the urgent need for gatekeeper responsibility, privacy-by-design principles, and robust industry standards to ensure these powerful technologies serve humanity without compromising our fundamental rights to privacy and security.