Redefining Trust and Authorization for the AI Era

• The unique challenges posed by AI, including deepfakes, fraud, and the emergence of AI agents disrupting traditional authentication and authorization frameworks.
• How modern identity technologies—such as identity graphs, adaptive authentication, and fraud prevention tools—can be leveraged to securely establish and scale trust.
• The role of emerging standards, compliance frameworks (like the AI Act), and advanced technologies in building future-proof solutions for AI identity management.

Security terminology often portraits a simple picture; good v. bad guys, inside v. outside, trusted v. zero-trust. This picture was never quite as simple even in the early days of the Internet, but with the advent of modern AI powered systems, this simplest picture is entirely obsolete and that means many of the technical building blocks of security and privacy as well. In this talk, Wenjing will discuss in non-technical terms why and how we should transition from security oriented language and mindset to a trust oriented one. He will further explore this theme and focus on the concepts and practices of privacy and trust. Trust is a very human concept. As we move from simple AI powered chats to reasoning and AI agents, privacy and trust will become critical limiting factors for the business and societal success of AI adoption. A new world populated with a lot of AI agents will be messy, it's time that we redefine our (now obsolete) trust framework to reflect such a new world reality.