The Lethal Trifecta Is Already on Your Laptops — Michael Patterson, Coder

Why watch

This talk explains Simon Willison's lethal trifecta security concept and why AI coding agents like Claude Code and Codex running locally pose significant risks. It covers how prompt injection transforms trusted data sources into attack vectors and presents three architectural guardrails: cloud development environments for isolation, model proxies for traffic logging and sensitive data stripping, and agent firewalls for command/domain control. Ideal for security engineers, platform engineers, and technical leaders evaluating AI coding agents in enterprise environments.