Your AI Agent Just Got Pwned A Security Engineer's Guide to Building Trustworthy Autonomous Systems
Автор: BSidesPhilly
Загружено: 2026-02-11
Просмотров: 4
Описание:
Presenter: Matt Maisel
A recent AI red teaming study by the UK AI Security Institute achieved a 100% attack success rate against all tested agents, with successful exploits in as few as 10 queries. If you're a security engineer building or securing agents, this should motivate you to act. As we give agents more autonomy, we expand the attack surface. Every autonomy level increases utility but also increases risk. Yet most teams ship agents with only basic prompt filtering. This talk delivers practical patterns for building secure agents. Attendees will learn what agentic systems are and why to use them, how each autonomy level creates new attack vectors, design patterns for agents, and guardrails that add security hooks to agent frameworks without breaking functionality. Attendees will leave with a security evaluation framework, code examples, and a pre-deployment checklist.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке:
