AI · SEMICONDUCTOR · FACT CHECK

SKD WIRE

판정 리포트 · 완료 2026-10-14 19:09

오픈AI, 추론 감시 어려운 '루프 트랜스포머' 적용 모델 안전 확보에 전력

대체로 사실

요약
14f1c85860fc4c/Hero_16x9.png?w=1600&h=900&fit=fill] 최종 판정: 대체로 사실
근거 — 직접 열람한 1차 출처 12건
  1. https://www.digitalfocus.news/news/articleView.html?idxno=24441
  2. https://www.theinformation.com/articles/secret-technique-behind-openais-astra-model-sparks-security-concerns
  3. https://aiweekly.co/alerts/openai-scrambles-to-fix-safety-issues-on-next-flagship-model
  4. https://techcrunch.com/2026/09/02/openais-new-reasoning-technique-alarms-ai-safety-experts/
  5. https://fortune.com/2026/09/03/reports-openais-astra-model-uses-a-new-more-efficient-ai-architecture-alarms-ai-safety-experts-who-worry-the-method-makes-models-harder-to-control/
  6. https://zdnet.co.kr/view/?no=20260904092159
  7. https://www.reddit.com/r/LocalLLaMA/comments/1wz00vv/
  8. https://magazine.sebastianraschka.com/p/gpt-6-astra-looped-transformers-and
  9. https://deploymentsafety.openai.com/gpt-6-astra
  10. https://openai.com/index/gpt-6-astra/
  11. https://openai.com/index/path-to-astra/
  12. https://deploymentsafety.openai.com/gpt-6-1-sol
이 판정 퍼가기 — 배지 임베드
대체로 사실
위 코드를 붙여넣으면 어디서든 이 판정 배지가 링크와 함께 노출됩니다.