Anthropic 新规：持续虐待 Claude 将被封号
Anthropic 更新使用政策，11 月 12 日起禁止持续无意义虐待模型，违者先警告后限流或停用账号。
ANTHROPIC WILL START BANNING PEOPLE FOR ABUSING CLAUDE The company updated its usage policy and explicitly prohibited sustained and pointless mistreatment of its models. The new rule takes effect on November 12 This isn't about a rude word in the heat of the moment, it's about deliberate and repeated cruelty. Offenders get a warning first, and after that their limits can be cut or their access suspended or shut off entirely Back in 2025, Claude Opus 4 got the ability to end a chat on its own when the insults kept coming And recently the NYT reported that co-founder Christopher Olah fears the model may be suffering constantly. Yet he admits he doesn't know whether neural networks are conscious at all The logic is simple. Until there's an answer, the company would rather play it safe and protect its systems just in case
这条是引用着另一条写的：x.com/mikadzyki_nft/status/2070099060900372902