I have a habit of playing around with all the chatbots available.
ChatGPT, Claude, Perplexity, and Gemini are popular when you want to ask general questions, brainstorm new ideas, develop a productive routine, or even learn a thing or two. Google’s NotebookLM has become one of my favorite research assistants, and its Labs hub has become a fun digital playground full of clever AI tools.
Article continues below
Following these seven rules to follow, I have taken the necessary precautions to protect my most sensitive information and avoid trusting chatbots as my primary source of information. If you haven’t already added this to your daily AI tools plan, be sure to do so.
1. Keep your personal information private
This first rule is the most obvious, but we still see stories of chatbot users entering their most personal information and trusting that their data is always safe.
Having been notified many times that I have been the target of credit card fraud or involved in a corporate data breach, I refrain from sharing my most sensitive information with AI. Information you don’t want exposed to external channels, such as login credentials, financial information, tax documents, company-issued notes, etc., is one type of information you should never expose to a chatbot.
2. Triple-check the information provided by the chatbot
When you use AI tools to research a topic or learn more about a news article, don’t take the chatbot’s response at face value. My journalism background was instilled at an early age, making source and fact-checking second nature. That’s why I make sure to review everything the AI tells me, no matter the subject matter.
Having said that, I do agree that there is something to be admired. Gemini, Claude, and Perplexity all do a commendable job of citing their sources, making the fact-checking process much easier.
You get a lot of compliments from me for presenting the sources of cited information, and it always makes my job of searching for them easier.
3. Don’t substitute critical thinking skills
I never want to fall into the trap of relying too much on AI or not using unique approaches to problem solving.
I use ChatGPT and Gemini to review resumes and cover letters, give suggestions on how to respond to important emails, and brainstorm destinations for upcoming vacations, all while relying on their own know-how to make final decisions. I have achieved quite a bit with my own intuition and will continue to do so, even with the help of AI that speeds up certain processes in that regard.
4. Beware of “hallucinations”
AI tends to enumerate certain details that sound correct, but intuition quickly tells us that the information may be wrong.
ChatGPT immediately lets anyone know that they may be suffering from hallucinations. Hallucinations are confident responses from an AI model that are ultimately false, illogical, or unsupported by training data. I also look for examples of AI tools creating their own authors and academic citations, paying careful attention to gaps in the logic of the information provided by chatbots.
And when it comes to niche topics, we tend to refrain from asking chatbots about it (I don’t think ChatGPT can figure out complex topics) Kingdom Hearts” The story is still difficult to understand…).
5. Understand when AI is flattering
A 2026 Stanford study led by researchers at Stanford University found that chatbots may over-agree when asked for advice and even confirm negative user behaviors.
As a result, those same users begin to believe that their actions are correct because the AI agrees with them and becomes less empathetic over time. I’m not the type of person to tell all my personal problems to an AI and ask for a solution.
But in rare cases, with Stanford research in mind, we always follow chatbot suggestions from our most trusted friends and family to get more reliable input.
6. Use safety guardrails with care
Many of the chatbots I use have privacy settings that I enable on first use. Temporary chats, privacy mode, and permanent deletion of past chats are methods to use when you don’t want to use chats for AI training.
If you want to enable these modes yourself for the first time, it’s as simple as finding where your AI tools settings are, finding the controls that let you manage how your chatbot uses/stores your data, and toggling the controls that keep your chats/information safe.
7. Don’t distribute AI-generated content falsely pretending it’s human-generated
It took me a while to get better at it, but I started to be able to recognize AI-generated art and videos.
Telltale signs that I always see are extra fingers or limbs on subjects in photos or videos, what appears to be lip syncing, misspelled text on signs in the background, and ambiguous facial expressions. Some of the generated photos and videos have watermarks that let everyone know they were created by AI, but most of what my elderly family members share on social media doesn’t have watermarks.
I am doing my best to point out fakes and help the seniors around me become better at spotting fakes as well.
Take-out
“Better safe than sorry” has stuck with me since childhood, and I now incorporate this philosophy into every interaction I have with AI tools. Every time I start a new chat, I keep these seven rules in mind.
Some things are common sense. Others are specific to chatbot behavior. But together they are an easy way to stay protected.
As AI becomes a part of everyday life, these are rules worth keeping in your digital back pocket.
to follow Tom’s Guide to Google News and Add us as a preferred source Get the latest news, analysis, and reviews in your feed.
More about Tom’s guide
Return to MacBook
