推文

@wangzhen1393032 · 2026-10-17 14:04

Anthropic’s Cyber Mission is a useful signal: frontier models are moving from demos into the security work nobody notices until it fails. The interesting question is not “can AI hack?” but “can defenders make verification continuous?”

曝光 5 · 评论 0 · 点赞 0 · 书签 0