They both outsourced the safety to each other, meaning there isn't any, and meanwhile the kids obviously are needing to talk to someone about their problems, left with something that is a huge risk for trusting at all.
OpenAI needs more user data, period. Only reason they would try this.