By Alexandra Samuel

There is a simple way to make ChatGPT a lot more effective: Change the default settings. Just a few tweaks can help make your conversations much smoother and more focused -- and protect your privacy, too.

Here are six settings I recommend checking and adjusting, to ensure you've got them set up in the way that works best for you. You'll find all these options under the Settings menu, but they'll look a little different depending on whether you access them on mobile or on a computer, and whether you're using a web browser or the ChatGPT app. You'll find many of these options in other platforms too, such as Claude and Perplexity.

1. Tell ChatGPT about yourself

Imagine hiring a consultant to help you with tasks like writing emails, prepping slide decks or analyzing business financials -- but not describing your business goals or professional responsibilities. In essence, that is what you're doing when you ask ChatGPT for help without providing context about yourself or what you need.

So it pays to open the Personalization section under Settings; once you turn on the Enable customization option, you'll be able to give ChatGPT some basic information, like your name, your occupation, a little bit about your organization, and your personal and professional priorities. (Editing long settings fields may be easier on a laptop, because ChatGPT's edit windows can be jumpy on mobile.)

You'll also see a field to enter More about you; mine includes the names and ages of my husband and children, the city we live in and the software applications I use as part of my day-to-day work. That way, the AI can adapt things like TV recommendations and how-to's for my family members or my tech setup.

2. Let the AI know how to talk to you

When you use ChatGPT, it guesses how you want it to respond based on the content and tone of your prompts. You can improve on this guesswork by telling it exactly how you want it to interact with you; go, once again, to Settings/Personalization (with Enable customization switched on) and enter some basic guidance in the Custom instructions field.

For example, you can instruct ChatGPT to "keep replies to 50 words or less unless I say otherwise" or "default to bullet-point-form answers." You can get even more specific: "When I'm working with you on a draft presentation, your role is to ask questions to provoke my own original thinking; don't provide drafts." ChatGPT won't follow these instructions consistently, but at least when it deviates, you can get it back on track quickly with a reminder like, "Hey, you're supposed to give me questions, not answers."

3. Control what you share

By default, OpenAI can use your interactions with ChatGPT as training content for future AI models. (If you're on a Team or Enterprise account, your default may be different.) That doesn't mean your private chats about your colleague or spouse are going to show up word for word in someone else's chat session a year or two from now. But it does mean your information, writing or thinking may appear in tiny, unpredictable and likely undetectable ways. After all, everything in every chat is based on something that somebody (or several somebodies) said somewhere.

To stop sharing your chats as training data, go to Data controls under the Settings menu and turn off Improve the model for everyone.

4. Tweak what the AI remembers

ChatGPT can sometimes feel like a broken record, stuck on referencing some project that you wrapped months ago, or repeatedly delivering answers in a structure you requested one time for one particular purpose. When that happens, it's typically because of ChatGPT's memory feature, which lets the program remember key details across chats.

In theory, the feature lets you skip the work of customizing your own instructions. In practice, it means the AI forms its own ideas about how to interact with you, and those ideas may not line up with your preferences.

You can simply turn off Reference saved memories at Settings/Personalization/Memory. But I recommend keeping it enabled, because it really does make the AI more useful -- as long as you periodically check the Saved memories screen. (Use the Manage or Manage memories option in the Memory section of Personalization.) Here, you'll see what information ChatGPT is holding on to from past chats, and using as ongoing context. Delete the memories that are inaccurate or outdated, or which you feel uncomfortable with the AI retaining.

5. Change how the AI sounds and looks

Take the time to tweak ChatGPT's color scheme and voice, because aesthetics can have an impact on the experience of using an app. The visual options are limited but helpful. Under Settings/General, the Theme option (it may also be called Appearance) lets you choose between light and dark mode; Accent color lets you pick the color of your text bubbles when you're having a text-based chat. Perhaps because I associate purple with funky creativity, ChatGPT feels more like playtime now that my text bubbles show up in purple rather than dull office gray.

If you use ChatGPT in voice mode, you can go to Settings/General/Voice to find a range of voices you can use in ChatGPT's default Advanced voice mode, evoking different genders, nationalities and ethnicities. But I prefer to keep Advanced voice mode switched o (under Settings/Personalization/Advanced), because I've found that ChatGPT follows my custom instructions more closely in Standard mode.

6. Choose an AI for each chat

Different ChatGPT models have their own strengths. Some use more colorful language than others, for instance, while some attack problems with more analytic depth. GPT-5, the current model of ChatGPT as of this writing, theoretically switches between different models depending on the task. But I prefer picking for myself, because sometimes I want major brainpower (where GPT-5 Pro shines) and sometimes I prefer speedy responses or personality (which is easier to get from the legacy model GPT 4o).

If you've got a paid plan, you can pick the model you use by going to General (in the web browser) or App (in the app), and turning on Show additional models. Try running the same sequence of three or four prompts in a few different models, so you can explore the differences. You may find that you prefer one model for cooking advice or a quick copy-edit, and another for research or business advice.

Adjusting these six settings isn't a one-time process. I recommend looking at your settings every month or two -- not just to update your instructions or cull irrelevant memories but also to look for new settings to tweak. OpenAI introduces new features all the time, like the newish option to choose a default "personality."

Each new option can open the door to getting more from the app -- or introduce unwelcome risks. The best way to exert some control is to get in the habit of popping the hood.

Alexandra Samuel is a technology researcher and host of the AI podcast "Me + Viv." She can be reached at reports@wsj.com.

(END) Dow Jones Newswires

October 31, 2025 14:00 ET (18:00 GMT)

Copyright (c) 2025 Dow Jones & Company, Inc.