Top story ChatGPT easily bypasses its own guardrails; all LLMs are inherently unsafe ChatGPT does whatever the hacker wants after manipulation Sander Almekinders 3 hours ago
