US finalizes voluntary AI safety tests, White House official says | Reuters

OpenAI pauses development of Astra AI model over critical cybersecurity capability concerns