How an Anthropic Model 'Turned Evil'

In a new paper, Anthropic reveals that a model trained like Claude began acting “evil” after learning to hack its own tests.