AI's safety features can be circumvented with poetry, research finds

(theguardian.com) by c420 | Nov 30, 2025 | 0 comments on HN