Redwood Research blog

The inaugural Redwood Research podcast

With Buck Shlegeris and Ryan Greenblatt

13 hrs ago • Buck Shlegeris

15

5

3:52:01

Recent LLMs can do 2-hop and 3-hop latent (no CoT) reasoning on natural facts

Recent AIs are much better at chaining together knowledge in a single forward pass

Jan 1 • Ryan Greenblatt

11

4

December 2025

Measuring no CoT math time horizon (single forward pass)

Opus 4.5 has around a 3.5 minute 50%-reliablity time horizon

Dec 26, 2025 • Ryan Greenblatt

9

1

Recent LLMs can use filler tokens or problem repeats to improve (no-CoT) math performance

AI can sometimes distribute cognition over many extra tokens

Dec 22, 2025 • Ryan Greenblatt

10

2

BashArena and Control Setting Design

We’ve just released BashArena, a new high-stakes control setting we think is a major improvement over the settings we’ve used in the past.

Dec 18, 2025 • Adam Kaufman and James Lucassen

8

The behavioral selection model for predicting AI motivations

The basic arguments about AI motivations in one causal graph

Dec 4, 2025 • Alex Mallen and Buck Shlegeris

13

4

November 2025

Will AI systems drift into misalignment?

A reason alignment could be hard

Nov 15, 2025 • Josh Clymer

15

2

1

What's up with Anthropic predicting AGI by early 2027?

I operationalize Anthropic's prediction of "powerful AI" and explain why I'm skeptical

Nov 3, 2025 • Ryan Greenblatt

20

2

October 2025

Sonnet 4.5's eval gaming seriously undermines alignment evals

And this seems caused by training on alignment evals.

Oct 30, 2025 • Alexa Pan and Ryan Greenblatt

26

5

2

Should AI Developers Remove Discussion of AI Misalignment from AI Training Data?

There is some concern that training AI systems on content predicting AI misalignment will hyperstition AI systems into misalignment.

Oct 23, 2025 • Alek Westover

9

3

Is 90% of code at Anthropic being written by AIs?

I'm skeptical that Dario's prediction of AIs writing 90% of code in 3-6 months has come true

Oct 22, 2025 • Ryan Greenblatt

14

2

Reducing risk from scheming by studying trained-in scheming behavior

Can we study scheming by studying AIs trained to act like schemers?

Oct 16, 2025 • Ryan Greenblatt

5