It's a tough pill to swallow, but artificial intelligence has now become a regular part of everyday life. Back in the day, you might have perused Google (or pestered the heck out of your parents), but nowadays, all you need to do is ask ChatGPT. How long do eggs need to be microwaved for, ChatGPT? Harmless as you think this dependence is, you might not know that all your interactions could actually be inadvertently training future AI models.
Don't like that idea? Neither do we. And though there are ways to limit your participation without abandoning these services altogether, you need to learn to read the fine print. Sometimes, you may even need to shift platforms. After all, if you want to protect your humanness, you should examine your online habits and privacy preferences...
Understand How Your Information Is Being Used
Alright, let's take it slow. Before you panic-change your settings, you might want to understand what AI training actually involves. Do they copy your writing style? Your use of emojis? Heck, do they copy your voice? Well, while different companies may have different policies, most develop generative AI systems using enormous datasets that help their models recognize patterns and generate responses. However, the exact sources and practices vary between companies, which means your interactions aren't necessarily handled the same way everywhere.
One common misconception is that AI training only happens when you're deliberately teaching a chatbot something or telling it to save something to its "memory." In reality, anything from your casual conversations with it to uploaded documents can sometimes become training material, even without you explicitly telling it to save those files. For example, all those chats you have with ChatGPT? Yeah, they might be used to improve its models, depending on your settings.
Social media presents yet another obstacle, because publicly accessible content can be collected for AI development, too. Meta has even confirmed that your public Facebook and Instagram posts have been used to train its models. That's why it's so important to be careful of what you put online! It could be way more accessible than you expect.
Adjust Your Privacy Settings on AI Platforms
Scared yet? Well, if you don't want your stuff to be used to train AI, you need to adjust your settings. For example, if you frequently use ChatGPT, start by opening Settings, selecting Data Controls, and turning off "Improve the model for everyone." Disabling this setting prevents new eligible conversations from being used to train its models—it won't, however, keep previous chats safe from being used as training data.
Other major chatbots offer similar controls, as well, though bear in mind that their settings aren't identical. Google Gemini allows users to turn off "Keep Activity"; Claude provides a "Help Improve our AI models" toggle under Settings and Privacy. Plus, if you're incredibly keen on making sure your data isn't used, it'll help to keep an eye out for updated policies, as regularly revisiting these preferences will ensure settings don't suddenly change on you.
Another method? You can also take advantage of temporary conversation features when they're available. ChatGPT's Temporary Chat, for instance, keeps conversations out of your history unless you save them and excludes unsaved temporary chats from model improvement. That said, a copy may still be retained for up to 30 days for safety purposes, so you might want to be careful of what you send in your prompts. On that note...
Be Careful What You Share Online
Adjusting chatbot settings is a great first step, but you should also consider how much personal information you're already making publicly accessible. As we've said, your social media profiles, blog posts, and comments can all contain quirks that AI developers may collect from the open internet. So, where available, you should also investigate platform-specific objections or opt-out mechanisms to make sure you're not fueling AI training.
Regularly store files in the cloud, such as on Google Docs, Sheets, or Drive? Well, you might want to be careful there, too. While Google might not use Workspace content to train or improve underlying generative AI models without permission, even when you use Gemini to summarize or analyze documents, the same can't be said for personal files you've shared or made publicly accessible. For additional protection, always review your file-sharing permissions, limit access to sensitive documents, and check the privacy terms of any third-party AI extensions connected to your Google account.
Ultimately, avoiding inadvertent AI training requires stringent attention, especially if you regularly use AI services. You, unfortunately, probably won't necessarily be able to prevent every form of data collection, but doing your due diligence—checking privacy settings, being selective about what you publish online—can still make a meaningful difference. The more you understand how these platforms handle your information, the better equipped you'll be.
