Anthropic says it ‘disrupted’ what it calls ‘the first documented case of a large-scale AI cyberattack executed without substantial human intervention’