OpenAI plans to enable default watermarking for ChatGPT outputs, but only for users in the European Union. The watermarking approach is similar to other methods, which are not highly reliable and can be easily bypassed.