Breaking

Sakana AI's 'AI Scientist-v2' Automates the Full Research Cycle, Landing First Peer-Reviewed AI-Generated Paper

July 17, 2026 at 06:16 EDT

AI Agents
Research & Papers
Open Source

Sakana AI's fully autonomous research system, The AI Scientist-v2, generates hypotheses, runs experiments, analyzes data and writes up manuscripts on its own—and produced what its creators say is the first entirely AI-generated paper to pass peer review at a workshop of a major machine-learning conference.

Autonomous Science · Sakana AI

An AI Wrote a Research Paper — and Peer Review Never Knew

The AI Scientist-v2 runs the full arc of research on its own — hypothesis, experiments, analysis, write-up. One of its manuscripts passed peer review at a major ML conference workshop, with reviewers unaware no human authored it.

1 / 3

Autonomously generated papers submitted to an ICLR workshop — one accepted

20–30 min

Time to write up the manuscript; a full run spans several hours

Human authors — and reviewers didn't know it

Score cleared the acceptance bar

The accepted paper scored above the average human acceptance threshold for the workshop.

Human avg
threshold

AI Scientist-v2
paper

The end-to-end pipeline

Ideation

Hypotheses

→

Tree Search

Parallel experiments

→

Analysis

Figures & data

→

Write-up

Manuscript

A manager agent grows the search tree in stages, selecting the most promising paths — guided by a vision-language model that refines plots.

What impressed

Peer-review acceptance as a genuine milestone
Systematic, parallelized agentic tree search
Open code, openly available proof point

Open questions

Runs execute code + reach the web — must be sandboxed
Cost and the ceiling of autonomy unclear
Little independent real-world usage yet

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.