Tag: AI safety

What METR’s May 2026 report says on agent deception

METR’s May 19, 2026 frontier risk report is one of the clearest public snapshots yet of what top internal AI…