Breaking Models AI Security
- WHO IT IS FOR: AI security researchers, red teamers, and engineers specializing in LLM vulnerabilities and defensive engineering.
- WHAT IT IS ABOUT: An in-depth technical deep dive into the OWASP LLM01:2026 Prompt Injection standard and recent real-world exploits.
- FORMAT: A small-group technical discussion and peer-review session for presenting research and pressure-testing exploits.
Join the Bay Area AI Security Research Meetup for an intensive session focused on the evolving landscape of prompt injection. We will move beyond theoretical risks to analyze operational threats, including zero-click document exfiltration, agent-driven destructive actions, and supply-chain backdoors within MCP-connected ecosystems. By using the OWASP LLM01:2026 framework as our anchor, we will dissect the mechanics of direct and indirect injections, RAG corpus poisoning, and multimodal payloads.
This is a collaborative, practitioner-led session designed for those working on the front lines of AI safety and security. Whether you are developing new defense mechanisms or engineering complex exploit chains, this meetup provides a rigorous environment to pressure-test your findings. Participants are encouraged to bring active research, papers, or defensive strategies to share with a community of peers who will provide critical, technical feedback.