Webinar Description
Key Takeaways
- Examines security risks when AI agents powered by large language models execute real business actions such as financial transactions, data access and code modifications
- Analyzes real-world incidents involving McDonald’s, McKinsey and Hugging Face to illustrate emerging agentic threats
- Addresses limitations of current security tools including WAFs, API gateways and LLM guardrails in protecting the full agentic path
- Designed for CISOs, security architects, DevSecOps teams and IT leaders in organizations deploying AI agents
- Provides practical steps for inventorying and analyzing agents, MCP servers and APIs
About the Webinar
“Beyond the LLM: What recent AI incidents tell us about the shifting attack surface” is a 25-minute virtual webinar organized by Salt Security with speakers from Openform. The session focuses on how AI agents in production environments have changed the security landscape for organizations, particularly when these agents can execute consequential business operations through LLMs, MCP servers and APIs.
The Agentic Security Gap
The webinar addresses a specific blind spot in current security approaches: most LLM security tools concentrate on prompt injection and model-level threats while leaving broader agentic behaviors unmonitored. AI agents now reason, call tools and act autonomously, moving beyond simple chatbot functionality. This creates exposure when agents access sensitive records, trigger critical workflows or modify code without adequate runtime monitoring.
Incident Analysis and Practical Guidance
The session draws on case studies from major organizations to demonstrate how these threats manifest in practice. Speakers discuss the need for runtime agent behavior monitoring and security posture management, offering actionable frameworks for organizations to inventory their AI agents and associated infrastructure.
Who Should Attend
This webinar suits security professionals, product managers and technology consultants working in organizations that deploy or plan to deploy AI agents. It is particularly relevant for those in finance, healthcare and technology sectors where data sensitivity and operational security requirements are high.