11 Things You Should Avoid Using ChatGPT For

Chatgpt was released in 2022 and in the years that followed his revolutionized daily life for millions of people. From the organization of schedules to the rewriting of emails, it is in many things. However, this chatbot is not able to manage all aspects of your life or your business.

Large languages ​​like Chatgpt sometimes generate incorrect or obsolete information while seeing completely confident. It is not huge if you think about a party theme or write an email of practice. But with regard to sensitive fields such as money (in particular your taxes), health or legal problems, a bad answer can create serious problems.

Do not miss any of our impartial technological content and laboratory criticism. Add CNET as a favorite Google source.

This is why it is just as important to know when to avoid pussy cat as to know how to get the most out of it. To help you avoid the biggest traps, here are 11 specific situations where to turn to an AI chatbot could cause more harm than good.

(Disclosure: Ziff Davis, CNET’s parent company in April, filed a complaint against Chatgpt Maker Openai, alleging that it violated Ziff Davis Copyrights in the training and exploitation of its AI systems.)

1. Diagnosis of physical health problems

I definitely nourished catgpt my symptoms out of curiosity, but the returning answers can read as your worst nightmare. When you pore on potential diagnoses, you can go from dehydration and flu to a certain type of cancer. I have a bump on my chest and entered this information in Chatgpt. And now, he told me that I could have cancer. In fact, I have a lipoma, which is not cancerous and occurs in one person in 1,000. My approved doctor told me this.

I’m not saying that there are no good use of chatgpt for health: this can help you write questions for your next appointment, translate medical jargon and organize a chronology of symptoms so that you can prepare you better. And this could help make visits to the doctor less overwhelming. However, AI cannot order laboratories or examine you, and it certainly does not carry insurance against professional faults. Know your limits.

2. Take care of your mental health

Chatgpt can offer grounding techniques, of course, but it cannot drop the phone when you have real problems with your mental health. I know that some people use Chatgpt as a substitute therapist. Corin Cesaric de Cnet found it slightly useful to work through sorrow, as long as it kept its limits at the front. But as a person who has a very real and very human therapist, I can tell you that Chatgpt is still only a pale imitation at best, and at the worst incredibly risky.

Chatpgpt has no experience, cannot read your body language or your tone, and has no capacity for authentic empathy. He can only simulate it. An approved therapist operates under legal mandates and professional codes that protect you from damage. Chatgpt does not do so. His advice can bring back, ignore the red flags or unintentionally strengthen bias in his training data. Leave the work deeper – hard, disorderly human work – to a real human formed to manage it properly. If you or someone you love are in crisis, please compose 988 in the United States or your local hotline.

3. Make immediate security decisions

If your carbon monoxide alarm is starting to reach, please do not open the Chatppt and ask it if you are real in danger. I would go out first and ask questions later. Large languages ​​cannot feel gas, detect smoke or send an emergency team. In a crisis, every second that you spend is a second, you do not evacuate or not informing 911. Chatgpt can only work with the remains of information that you feed it, and in case of an emergency, it can be too little and too late. So, treat your chatbot as a post-incident explanator, never a first speaker.

4. Obtain personalized financial or tax planning

Chatgpt can explain what an ETF is, but he does not know your debt / income ratio, the state tax branch, deposit status, deductions, retirement objectives or risk appetite. Since his training data can stop in step from the current taxation year and the latest rate increases, his advice may well be expired when you press ENTER.

I have friends who throw their totals 1099 in Chatgpt for a return of DIY. The chatbot simply cannot replace a CPA which can take a hidden deduction worth a few hundred dollars or report an error that could cost you thousands of people. When real money, deposit deadlines and IRS penalties are at stake, call a professional, not an AI. Also be aware that everything you share with an IA chatbot will probably be one of its training data, and this includes your income, your social security number and your bank routing information.

5. Process confidential or regulated data

As a technological journalist, I see embargoes landing in my reception box every day, but I never thought of launching any of these press releases in Chatgpt to obtain a more in -depth summary or explanation. This is because if I did, this text would leave my control and land on a third-party server outside the railings of my non-circuit agreement.

The same risk applies to customer contracts, medical graphics or everything that is covered by California Consumer Privacy Act, HIPAA, the GDPR or an old commercial-secret law. It applies to your income taxes, your birth certificate, your driving license and your passport. Once sensitive information is in the prompt window, you cannot guarantee their storage, which can revise it internally or if it can be used to form future models. Chatgpt is also not immune to pirates and security threats. If you don’t stick it in a Slack public channel, don’t stick it in Chatgpt.

6. do something illegal

This one is explicit.

7. Cheat school work

I would lie if I said that I had never cheated on my exams. In high school, I used my first generation iPod touch to take a look at a few heavy equations that I had trouble memorizing in AP Calculation, a blow of which I am not particularly proud. But with AI, the magnitude of modern cheating makes this look remarkably tamed.

Turnitin and similar detectors improve to locate the prose generated by AI each semester, and the teachers can already hear “Chatgpt Voice” to a distance mile (thank you for spoiling my beloved dash). The suspension, expulsion and revocation of your license are real risks. It is best to use Chatgpt as a study boyfriend, not a ghost. You are also mistaken for an education if you have a chatpt doing the job for you.

8. Surveillance of information and news from the last

Since Openai deployed Chatgpt research at the end of 2024 (and opened it to everyone in February 2025), the chatbot can recover new web pages, stock market quotes, gas prices, sports scores and other numbers in real time when you ask, with clickable quotes so that you can check the source. However, it will not broadcast the continuous updates on its own. Each updating needs a new invitation, so when the speed is critical, the data flows live, the official press releases, the information sites, the push alerts and the streaming coverage are always your best bet.

9. Jamoun

I was lucky with Chatgpt and hits one by three in the NCAA male basketball championship, but I would never recommend it to anyone. I saw Chatgpt Hallucine and provide incorrect information on players’ statistics, poorly declared injuries and victory records. I only cashed because I checked each complaint against ratings in real time, and even then, I was lucky. Chatgpt cannot see the score of the box of tomorrow, so do not count on it to get this victory.

10. Drafting of a will or other legally binding contract

Chatgpt is ideal for decomposing basic concepts. If you want to know more about a revocable living trust, ask yourself. However, when you ask him to write real legal text, you launch the dice. The rules of succession and family vary according to the states, and sometimes even by county, therefore jump a signature of witness or omit the notarization clause can ensure that your document is launched. Rather, let Chatgpt help you build a question list of questions to your lawyer, and then pay this lawyer to transform this control list into a document being held before the court.

11. Make art

This is not an objective truth, just my own opinion, but I don’t think AI should be used to create art. I am in no way of anti-artificial intelligence. I use Chatgpt to think about new ideas and help my titles, but it is supplementation, not a substitution. By all means, use Chatgpt, but please do not use it to make the art that you then pass like yours. It’s a bit disgusting.