AI Security After Codex and Claude Code — Zico Kolter & Matt Fredrikson, Gray Swan
Source
youtube.com
Author
Latent Space
Date
Why it matters
If automated attackers beat human red teamers, manual red-team passes understate risk for deployed agents. Shows where AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → security evalA repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.Full definition → is heading.
Terms in this piece · Glossary
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
eval — A repeatable test for AI quality — a set of tasks plus scoring — used the way software teams use test suites, because model output is too variable to judge by eyeballing.
videoInside xAI: Building Grok Imagine in 3 Months, Videogen vs World Models, and Video Agents— Ethan He
videoBest of 2024 in Agents (from #1 on SWE-Bench Full, Prof. Graham Neubig of OpenHands/AllHands)
videoTrillion Token Context. No, Really — Anima Anandkumar & Benedikt Jenik, Accelerated Understanding
videoThe Unreasonable Effectiveness of Reasoning Distillation: using DeepSeek R1 to beat OpenAI o1
