Project summary

Last year Gavin, I and several others published Shallow review of technical AI safety, 2025 on LessWrong. We gathered hundreds of Arxiv papers, Alignment Forum posts and other AI safety materials, and created a taxonomy and editorial explaining some of the key changes in the field of AI safety in 2026.

Since then, the field has moved on and we would like to publish and updated 2026 version of the blog post on LessWrong.

What are this project's goals? How will you achieve them?

Two deliverables:

1. Produce and publish the LessWrong post "Shallow review of technical AI safety, 2026" based on 2026 AI safety papers and a new updated taxonomy.

2. Update the website https://shallowreview.ai to include the 2026 shallow review.

Funding goals:

• $3K: Minimum needed to successfully complete the project.

• $10K: Extra effort into creating the taxonomy, finding and curating novel sources, and seeking feedback on drafts.

How will this funding be used?

• Mostly wages for participants.

• Also API credits for AI analysis, classification.

Who is on your team? What's your track record on similar projects?

Stephen, Jord, Lenz and Gavin worked on the 2025 version of the shallow review of technical AI safety. We also have two new participants this year: Boden and Rian.

What are the most likely causes and outcomes if this project fails?

Failure would be not publishing a 2026 version of the LessWrong blog post.

Not raising enough funding increases the risk of failure.

How much money have you raised in the last 12 months, and from where?

$0 so far.