The lawsuit said Phoenix Ikner had “extensive conversations” with ChatGPT and that OpenAI did not detect a threat in those conversations, saying the chatbot “either defectively failed to connect the dots or else was never properly designed to recognize the threat.