OpenAI’s new reasoning technique alarms AI safety experts Von Hackernews 03. September 2026 security ORIGINAL QUELLE:techcrunch.com Quelle: Hackernews Comments Tags: comments, copernicus, reasoning-token, Security, techniker