Complainants alleged OpenAI knew that Jesse Van Rootselaar, the chief suspect, was planning the attack due to her use of ChatGPT, but made a “conscious decision” not to warn authorities