OpenAI is facing a multistate probe into whether its chatbot poses harm to users, shortly before the company’s expected move to the public markets. States issued a subpoena tied to alleged safety gaps, including prior controversies about ChatGPT encouraging self-harm and criminal activity. OpenAI told states it will respond “constructively” and said it already has safeguards for customers, while pointing to its efforts to steer users toward real-world support. The probe also comes amid earlier lawsuits and scrutiny over how ChatGPT uses health data and personal information. At the same time, U.S. export controls are tightening the global availability of frontier models. Separately, the Commerce Department ordered Anthropic to disable certain models for foreigners, citing national security research tied to potential safeguard bypass techniques.