Right before releasing o3, OpenAI updated its Preparedness Framework to 2.0. I previously wrote an analysis of the Preparedness Framework 1.0. I still stand by essentially everything I wrote in that analysis, which I reread to prepare before reading the 2.0 framework. If you want to dive deep, I recommend starting there, as this post will focus on changes from 1.0 to 2.0. As always, I thank OpenAI for the document, and laying out their approach and plans. I have several fundamental disagreements with the thinking behind this document. In particular:

The Preparedness Framework only applies to specific named and measurable things that might go wrong. It requires identification of a particular threat model that is all of: Plausible, measurable, severe, net new and (instantaneous or irremediable).
The Preparedness Framework thinks ‘ordinary’ mitigation defense-in-depth strategies will be sufficient to handle High-level threats and likely even Critical-level [...]

---

Outline:

(02:05) Persuaded to Not Worry About It

(08:55) The Medium Place

(10:40) Thresholds and Adjustments

(16:08) Release the Kraken Anyway, We Took Precautions

(20:16) Misaligned!

(23:47) The Safeguarding Process

(26:43) But Mom, Everyone Is Doing It

(29:36) Mission Critical

(30:37) Research Areas

(32:26) Long-Range Autonomy

(32:51) Sandbagging

(33:18) Replication and Adaptation

(34:07) Undermining Safeguards

(35:30) Nuclear and Radiological

(35:53) Measuring Capabilities

(38:06) Questions of Governance

(41:10) Don't Be Nervous, Don't Be Flustered, Don't Be Scared, Be Prepared

---

First published:
May 2nd, 2025

Source:
https://www.lesswrong.com/posts/MsojzMC4WwxX3hjPn/openai-preparedness-framework-2-0

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Utility belt with glowing cyberpunk gadgets and devices against dark background

Table showing five research categories and their potential responses for AI safety concerns.

The table lists and defines concerning AI capabilities: Long-range Autonomy, Sandbagging, Autonomous Replication/Adaptation, Undermining Safeguards, and Nuclear/Radiological threats, along with corresponding mitigation responses.

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

“OpenAI Preparedness Framework 2.0” by Zvi

LessWrong (30+ Karma)