As of my last update in September 2021, fine-tuning ChatGPT on custom data is not directly supported. OpenAI provides access to the base models for fine-tuning, but the specific models like ChatGPT cannot be fine-tuned on custom data due to certain technical limitations and concerns about potential misuse.

If you’re interested in customizing ChatGPT’s behavior, you can use prompts and instructions to guide its responses, but direct fine-tuning on custom data is not available for this model.

It’s recommended to review OpenAI’s official documentation and updates to see if there have been any changes to their policies or offerings related to fine-tuning on custom data since my last knowledge update. Always ensure that you are following the most up-to-date and accurate information from OpenAI.