In brief 13 November
Anthropic advises that, to keep Fable 5.2 responsible for an action, you should write “ask for approval…
Anthropic advises that, to keep Fable 5.2 responsible for an action, you should write “ask for approval before deleting files” instead of “never delete files”, and reply with the specific instruction rather than a simple “go ahead”. It follows the guide the paper reported on.