High-stakes Sabotage: Threat Model and Mitigations

The Don’t Worry About the Vase Podcast is a listener-supported podcast. To receive new posts and support the cost of creation, consider becoming a free or paid subscriber.

* 00:00:00 - Introduction

* 00:00:40 - How RSP v3.0 Works

* 00:33:42 - You Came Here For An Argument

* 00:37:56 - The Problem Remains Unsolved

* 00:41:53 - Wow That Thing We Did Was Pretty Risky, Huh?

* 00:42:55 - Risk Report Number 1

* 00:44:57 - Listen All Y’all It’s Sabotage

* 00:58:19 - Looking Forward

* 01:00:07 - Claude Gov

* 01:00:26 - What Is A Strong Argument?

* 01:01:42 - Recursive Self-Improvement

* 01:04:28 - Non-Novel Chemical and Biological Weapons

* 01:06:55 - Novel Chemical and Biological Weapons

* 01:07:47 - Cross-Cutting Content (Section 6)

* 01:11:14 - Risk Report Report

https://open.substack.com/pub/thezvi/p/anthropic-responsible-scaling-policy-46a?utm_campaign=post-expanded-share&utm_medium=web