Think Before You Chat: Why Sharing Personal Info with AI Could Haunt You

Credits: Shutterstock

AI chatbots such as ChatGPT are able to hold conversations about nearly any subject with you, which gives the sensation of speaking to a friendly chat companion. But don’t be misled — anything you post can be saved and potentially come back to haunt you. Unlike a toaster, an AI chatbot retains what you tell it, so sharing personal details is a gamble of sorts.

How Chatbots Use Your Data

The companies that make AI chatbots use your data to train their models. It’s kind of like a scene in a movie where a character learns new phrases — but instead of comedic one-liners, these bots are learning from your personal info. OpenAI, for example, states plainly in its terms that it can use your input to improve its models. Unless you go into settings and turn off chat history, everything you type is fair game, including passwords, addresses, and files.

What Happens to Your Data?

ChatGPT’s terms say it may “aggregate or de-identify Personal Information,” which sounds harmless, but it leaves the door open for your data to be used in ways you might not expect. Even if a chatbot isn’t designed to leak information, things can still go wrong.

The Risks of Data Breaches

AI companies may not intend to misuse your data, but that doesn’t mean it’s completely safe. In May 2023, hackers took advantage of a flaw in ChatGPT’s system, gaining access to personal data from chat histories. Stolen details included names, social security numbers, job titles, and even contact information. OpenAI fixed the problem, but this did not help the 101,000 users whose data was already out there.

Companies Are Taking Action

Businesses have also been caught off guard. Samsung engineers accidentally leaked sensitive source code to ChatGPT, leading the company to ban its use at work. Other companies, including Bank of America and JPMorgan, have done the same to prevent leaks of confidential information.

Is Change Coming?

Governments and industries are starting to take AI security seriously. In October 2023, President Joe Biden signed an order emphasizing that AI systems should protect privacy. However, the U.S. still lacks clear laws preventing AI companies from using data without consent. Until stricter regulations are in place, AI chatbots can still learn from what you share.

How to Stay Safe

Until privacy laws catch up, your best bet is to be cautious. Chatbots may seem friendly, but they’re still algorithms, not trusted advisors. No matter how much they flatter you, it’s best to keep personal details to yourself.