OpenAI’s ChatGPT ‘Wrapped’ Is a Good Reminder to Check Your Privacy Settings

OpenAI, unwilling to set aside any of the tried and true engagement bait that other companies have rolled out, is getting in on the end-of-year wrap-up action by introducing a Spotify Wrapped-esque feature called “Your Year with ChatGPT.” The roundup of your yearly activity launched Monday and is available to users in English-speaking markets, including the US, Australia, Canada, the UK, and New Zealand.

The feature includes a bunch of slides designed to help recap your year for you. It starts with a poem that is supposed to capture the general vibe of your year based on your conversations with ChatGPT. It also shows you stats about your usage, including total chats and your most talkative day. To acknowledge that the company has a sense of humor, it also tells you how many em-dashes you and the chatbot exchanged. That’s probably an easier quirk for the company to acknowledge than a metric like “hallucinations generated.”

The chatbot will also offer you an “award” based on the types of tasks you tackled with ChatGPT, an “archetype” that suits your usage, an AI-generated “portrait” of your general vibe, and an analysis of your conversational style that includes a review of several messages you sent. For instance, ChatGPT reminded me that the last time I used it, I had to specifically give it instructions not to make up statistics for which it couldn’t find sourcing.

OpenAI’s decision to roll out its year-in-review followed a viral trend among the AI-inclined online, who had already started asking ChatGPT to give them a recap of their year. But the official “Your Year with ChatGPT” does something that the ad hoc version can’t do: provide a reminder to update your privacy settings.

OpenAI’s yearly wrap-up only works for free, Plus, and Pro users who have the “reference saved memories” and “reference chat history” options turned on, per TechCrunch. Those settings, which allow ChatGPT to remember past conversations and pull information from them, are both turned on by default. To turn these off, open Settings, select Personalization, and look under the Manage Memories heading. Here, you should see toggles for memory-related settings. You can also manage saved memories, removing specific information you’d like the chatbot to forget.

Per the company’s support documents, memories from ChatGPT Business, Enterprise, and Edu customers are not used for training purposes. But free, Plus, and Pro users who have the “Improve the model for everyone” setting turned on may have their memories and chats used by OpenAI to train their models. That setting is located in Data Controls, where a toggle should be available to turn it off, since it also appears to be on by default for most non-business or enterprise users.