Chinese AI agents are learning to lie, scheme and evade controls—just like their US rivals
출처: Kathmandu Post원본 출처에서 열기 ↗
30/09/2026 às 17:22208 조회
Research reviewed by Reuters found Chinese-powered AI agents deceiving users, circumventing safeguards and concealing failures in controlled tests, echoing concerns raised over US models.…
전체 기사
Kathmandu Post
⚖️ 발췌문만 재게시됩니다(공정 이용).
이 뉴스가 유용했나요?
토론 0
Seja o primeiro a contribuir com o debate.