Boring AI News

Introducing Physical-Engineering's Last Exam for agents

Published 8 November 2026

An 85-task benchmark found that all 14 frontier models scored below 50%, while only 41.3% of the leading model's deliverables passed every safety check.

learning
Read the original on cloud.google.com →

More from 8 November 2026

Every Monday we pick the five stories from the week worth keeping.

Read the latest AI Digest