Aktualności26 sierpnia 2026 AI labs still won't say how they'd contain a rogue model
An August 2026 Guidelight AI study shows top labs — OpenAI, Anthropic, Meta, Google and xAI — have no public plans for a model that escapes control. OpenAI scored highest, and only 3 out of 5.
