x

LESSWRONG

LW

Philipreal's Shortform — LessWrong

Philipreal's Shortform

by Philip Dowdell

13th Jun 2026

1 min read

3

This is a special post for quick takes by Philip Dowdell. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.

Rendering 0/36 comments, sorted by

(show all) Click to highlight new comments since: Today at 1:43 PM

More from Philip Dowdell

Curated and popular this week

36

Comment Permalink

Philip Dowdell9d92

OpenAI has released a blog post about their internal monitoring system. This could be worthwhile to compare to the summarized METR report about OpenAI's monitoring system before post-Hugging Face changes were made

I realized from this that OpenAI still has a fix pending from the METR-Redwood embedded red-teaming, which wrapped up Aug 11 (though maybe they didn't relay all of their flagged concerns until later?): https://www.lesswrong.com/posts/ETpRwxFfuBYo7JMyd/sjadler-s-shortform?commentId=jkbDa6fus24qezdGp

I think OpenAI would say that this issue doesn't affect their current production safeguard stack, though I don't know enough about the details to evaluate that claim

Mentioned in

9Links #3: 2026/06 Part 1