OpenAI Publishes Safety Card for GPT-6 Sol and GPT-6 Luna Rolling Out in ChatGPT

OpenAI's deployment safety documentation for GPT-6 Sol and GPT-6 Luna, which are replacing GPT-5.6 versions in ChatGPT globally, shows stronger jailbreak and prompt-injection resistance alongside statistically significant regressions in self-harm and age-restricted content handling.