“OpenAI is working on research for content provenance and authenticity tools, such as watermarking, to help people identify deepfakes and misinformation.”