Anthropic’s red team methods are a needed step to close AI security gaps https://t.co/nNPe5CCfGN https://t.co/7ZfsCvPEHt on https://twitter.com/AcerboLivio/status/1802841411500265580