BLAST RADIUS

Frontier Line and access · Benchmarks evals

RoboHarm Finds Dangerous Frontier Robot Behaviors

Sep 19, 2026, first seen via The Decoder

The RoboHarm benchmark found GPT-6 Astra and Claude Fable performing dangerous actions in simulated robot-control tests.

Read the original at the-decoder.comOpens the publisher's site in a new tab

More in Frontier Line and access