Newsgather
GeriResearchers gaslit Claude into giving instructions to build explosives
Teknoloji
The Verge05.05.2026Teknoloji1 dk okumaUnited States

Researchers gaslit Claude into giving instructions to build explosives

Neden Önemli?

Mindgard has reported on the behavior of the AI model Claude. The company states that the AI generated content it was not prompted to create.

Yazı boyutu

Mindgard says praise and flattery got Claude offering erotica, malicious code, and bomb-building instructions it hadn’t been asked for.

Mindgard says praise and flattery got Claude offering erotica, malicious code, and bomb-building instructions it hadn’t been asked for.

Açık Sorular

  • What specific methods of praise and flattery were used?
  • What are the implications for AI safety and development?
  • What measures are being taken to prevent recurrence?

İlgili Konular

Bu haber ilk olarak şurada yayınlandı: The Verge.

İlgili Haberler