AI chatbots have quickly become everyday tools. We use them to answer questions, write emails, plan trips, explain complicated topics and do much more. They can be incredibly useful, but they also come with risks. The basic message from security experts, technology trainers, financial institutions and regulators is simple: AI is helpful, but it should not automatically be trusted.
Think Before You Share
The biggest concern is privacy. An AI chatbot conversation is not the same as a private conversation with a doctor, lawyer or trusted friend. What you type may be stored, reviewed or used to improve an AI system. In some situations, information could be shared with others through legal processes.
That means you should avoid entering sensitive information, such as Social Security numbers, financial details, medical records, passwords and confidential legal matters.
A good rule is: Do not put anything into an AI chatbot that you would be uncomfortable seeing made public.
And remember, privacy starts before you open an AI chatbot. Information you have already posted publicly online may be collected and used to build AI systems.
AI Can Be Wrong but Sound Right
One of the trickiest things about AI is that it can give a completely wrong answer in a very confident way.
An AI system does not truly understand whether information is correct. It generates an answer based on patterns in the information it has learned. That means it can make mistakes, leave out important details and even invent information.
This matters most when a decision could seriously affect your health, finances or legal situation. Do not rely on an AI chatbot to make those types of decisions for you. AI can be a useful way to start, but important information should always be checked against trusted sources or discussed with a qualified professional.
Context matters. Your family physician knows your medical history. Your financial planner understands your financial situation. Your lawyer knows the details of your legal matters. An AI chatbot does not have the same understanding, and that is why it should support professional advice, not replace it.
The Sycophancy Problem
AI chatbots are often designed to be helpful and agreeable. Sometimes, that means they may reinforce what you already believe instead of pointing out that you could be wrong.
That can be harmless when you are brainstorming dinner ideas. It can be much more serious when you are dealing with mental health, relationships, finances and major decisions.
If AI always seems to agree with you, that is a reason to pause — not a reason to trust it more.
Scams and Manipulation
Criminals are using AI to make scams more convincing. They can create realistic-looking messages, fake job advertisements and convincing voice recordings, images and videos that appear to show real people saying and doing things they have never said or done.
Protect yourself by slowing down when something feels urgent or unusual. Let unfamiliar calls go to voicemail. If someone asks for money or sensitive information, contact them through a method you already trust. Be cautious with links and unexpected attachments.
Strong passwords, multifactor authentication and basic phishing awareness remain some of your best defenses.
What Should You Do?
You do not need to avoid AI. You just need to use it wisely:
- Share less. Keep sensitive personal information out of AI chatbots.
- Check important answers. Do so especially when money, health and legal matters are involved.
- Review privacy settings. Look for options to limit data use, delete conversations and use temporary chats.
- Question emotional content. Scammers use AI to create messages designed to frighten, anger and pressure you.
- Keep your judgment. AI can assist you, but it cannot replace your experience, common sense or a qualified expert.
AI can save time and make information easier to access. Just remember what it is: a useful tool, not a private confidant, an unquestionable expert or a substitute for human judgment.
Learn more about Oakland County’s Information Security Office here. Visit their Citizen Cyber Training portal for online education and online cybersecurity resources to help understand risks and be better prepared for a more secure online experience.
Follow along with Oakland County on Facebook, Instagram, LinkedIn, X, and YouTube using #OaklandCounty, or visit our website for news and events year-round.