Observed facts: Google Research says Deep Review is launching publicly. The company reports that it was deployed across more than 15,000 submissions at STOC, ICML, NeurIPS and ICLR, and describes it as a system for analyzing technical documents. These are company-reported facts, not an independent evaluation.
Mechanism: Deep Review applies AI to technical-document analysis in conference-associated workflows; public launch could extend access beyond prior deployments.
Why now: The public launch is a concrete availability change, and the reported use history provides a deployment signal.
First implication: Research teams can test whether it reduces document-analysis effort, while retaining human review.
Evidence boundary: The source provides Google's launch statement and deployment count, but no independent benchmark, accuracy results, measured time savings, or evidence of effects on decisions.
First watch condition: Watch for public access details and independent evaluation of error rates and human review effort on representative technical documents.
The story starts a distinct analytical record because no prior New Runtime record is supplied; the related historical context is not enough to infer an existing analysis of this launch.