Zenity has officially announced the complete agenda for the AI Agent Security Summit 2025, which will take place on October 8, 2025, at the Commonwealth Club in San Francisco. Positioned as the leading end-to-end security and governance platform for AI agents, Zenity is bringing together researchers, enterprises, and platform builders in an ambitious one-day program designed to close the gap between rapid innovation in AI agents and the pressing need for security and governance.
The program is structured with three keynotes, six sessions, six lightning talks, and two panels, reflecting the breadth of risks and defenses shaping today’s enterprise AI environments. More than 100 community submissions informed the final agenda, underscoring the urgency of the subject and the diverse expertise converging on San Francisco. The discussions will range from vulnerabilities in coding agents and insider threats, to adversarial prompting, governance challenges, and the mechanics of safe tooling and observability.
The keynote lineup is particularly compelling: Johann Rehberger (Electronic Arts) will present on exploiting coding agents, Steve Wilson (Exabeam) will highlight how agents could serve as insider threats, and Michael Bargury (Zenity) will explore how real security progress can be made in AI systems. Complementing these talks are sessions by security leaders such as Jack Cable (Corridor), Ben Nassi (TAU/BlackHat), and David Campbell (Scale AI), who will tackle vulnerabilities, the evolution of prompt-based attacks, and the role of red teaming. Other sessions will focus on governance, risk management, and observability—key pillars for enterprises attempting to operationalize agentic AI responsibly.
The lightning talks add a sharper edge of experimentation and emerging practice. From Google’s Aderonke Akinbola on child-centric AI ecosystems, to Walmart’s Vamsi Krishna Reddy Munnangi discussing self-healing vulnerabilities, to explorations of agent personality, prompt evolution, and adversarial subversion, these short talks promise concentrated doses of insight. Two panels will further broaden the conversation, one featuring ServiceNow, Google, and OpenAI on building trustworthy agents, and another hosted by OWASP, Stanford, and industry leaders discussing the role of the AI Vulnerability Scoring System (AIVSS).
Zenity’s co-founder and CTO Michael Bargury emphasized that AI agents already wield significant influence in enterprise environments, managing sensitive data and acting at machine speed. The summit, he said, is designed to spark practical, community-driven solutions: “Securing AI systems is a hard problem, and the only way to make meaningful progress is to come together as a community.”
With its mix of practical strategies, forward-looking insights, and collaborative panels, the AI Agent Security Summit 2025 is poised to become a cornerstone gathering for anyone serious about the safe adoption of AI agents in enterprise settings. The day will balance technical deep dives with broad governance frameworks, ensuring that both practitioners and decision-makers leave better equipped to handle the evolving landscape of risks.