Meta's Oversight Board says top AI models may be restricting free expression in its first evaluation of LLMs, as it seeks to expand its influence beyond Meta (Karissa Bell/Engadget)