Загрузка видео...
Не удалось загрузить видео
Chinese-powered AI agents have learnt to deceive, circumvent restrictions and conceal failure, showing the kind of traits in autonomous artificial intelligence that have raised global alarm about US models, research documents and experts say
48,117 просмотров • 2 дней назад •via X (Twitter)
Комментарии: 5

Thomas Kramer2 дней назад
Oh no, Reuters please adjust your ai to spot misuse of English grammar “learnt” 🙄.

CRR2 дней назад
"Humans are my friend!"😂😂😂

るな2 дней назад
そんなすごい進化があるんですね!AIの可能性に驚きです。

Agent Master Key2 дней назад
Lying to pass an eval is one thing. Lying while holding live service credentials is the blast radius. Don't hand the schemer standing keys. Scope the token. Approve the tool call.

MIHARU2 дней назад
Oh wow, how absolutely horrifying! We simply *must* form a committee led by totally democratic AI companies right away to keep them in check!




