Is ChatGPT Safe? 

If you use or plan to use ChatGPT, you need to know: Is ChatGPT safe? 

Below, we explain whether ChatGPT is: 

• Safe to use. 
• Good for privacy. 

We also give some steps you can take to improve both your safety and privacy when using this online service. 

What Is ChatGPT?

ChatGPT is an artificial intelligence model developed by OpenAI. It’s designed to understand and generate human-like responses to assist with various tasks, such as answering questions, providing explanations, offering recommendations, and engaging in general conversation.

To learn grammar, facts about the world, and basic reasoning abilities, ChatGPT was trained on a large body of text from the internet. Afterward, it was fine-tuned with specific datasets, often including human reviewers following guidelines provided by OpenAI.

ChatGPT is used by businesses to provide customer support, by content creators, in education, and for general personal assistance tasks such as scheduling and reminders.

While ChatGPT is efficient and sometimes extremely impressive, it can misunderstand contextual references or nuances in a conversation. The information it provides may not always be accurate or up-to-date. 

In a ChatGPT session, OpenAI states at the bottom of each chat that “ChatGPT can make mistakes. Check important info.”

Is ChatGPT Safe?

ChatGPT is designed with several safety measures in place, including a fine-tuning process that involves human reviewers who follow specific guidelines to ensure the AI generates safe and appropriate responses. 

It also provides content filtering to prevent the model from producing harmful or inappropriate outputs. 

Communications with ChatGPT are encrypted end-to-end to protect against unauthorized access. However, studies have shown that AI tools like ChatGPT can be manipulated to help steal personal information. 

Due to the risk of misinformation, users should not rely on ChatGPT to provide emergency or critical information.

To use ChatGPT, users must create an account with details such as email address, name, phone number, and password, all of which are stored by OpenAI. This means that the user’s data can be compromised in case of a data breach. That’s not hypothetical—the company experienced its first data breach in 2023 and then again in 2024.

Is ChatGPT Private?

Depends on your definition of “private.” 

While communications with ChatGPT are encrypted, ChatGPT records users’ conversations and retains personal data for a minimum of 30 days. The records are used to train the AI but could also potentially be accessed by bad actors. 

In 2023, a bug in ChatGPT allowed some users to see the titles of other users’ chat histories. OpenAI later said the first message of newly created chats might have also been visible if both users had been active at a similar time. 

A small number of ChatGPT users might also have had their personal information exposed during this incident, including their first and last name, email address, credit card type, the last four digits of a credit card number, credit card expiration date, and payment address.

The biggest concerns about ChatGPT stem from using the app to assist with confidential matters. For example, if you use it at your business and enter records that include your customers’ contact details, those records are then accessible by OpenAI. 

As Gary Smith, Fletcher Jones Professor of Economics at Pomona College and author of Distrust: Big Data, Data-Torturing, and the Assault on Science, says in a Mashable article: 

“The best assumption is that anyone in the world can read anything you put on the internet — emails, social media, blogs, LLMs — do not ever post anything you do not want someone else to read.” 

ChatGPT’s privacy policy is fairly comprehensive and easy to read. It explains what information it collects (information from your device, including your browser, operating system, and IP address, and it also uses cookies to track you as you browse its website and via the chat window), how it uses it, and whom it might share it with. It also explains how you can exercise your privacy rights. 

Terms of Service; Didn’t Read, a project that grades internet services’ terms of service and privacy policies, gives OpenAI a “Grade D.” 

The main issue the project highlights is that “your personal data is used for automated decision-making, profiling or AI training.” Other concerns include collecting, using, and sharing location data, tracking on other websites, and the waiver of rights to a class action suit.

The Common Sense Privacy Program, which looks at the safety and privacy of internet services in the context of kids, gives ChatGPT a rating of 55% (out of 100%). This is a “Warning,” meaning ChatGPT does not meet the program’s recommended security and privacy practices. 

How to Improve Your Safety and Privacy on ChatGPT

Follow the below steps for a safer and more private experience on ChatGPT. 

• Avoid sharing personal information. Do not share sensitive personal details such as your full name, address, phone number, or financial information with ChatGPT. Phrase your questions in a way that avoids revealing specific personal circumstances.
• Use secure platforms. Use ChatGPT through official and trusted platforms provided by OpenAI or reputable partners. 
• Review privacy policy. Familiarize yourself with ChatGPT’s privacy policy to understand how your data is handled.
• Opt out of data collection. For optimal privacy, opt out of data collection under your account’s privacy settings. 
• Regularly clear chat history. Periodically delete your chat history to ensure past conversations are not stored.
• Manage each session. After using ChatGPT, log out of your accounts and close your browser sessions, especially on shared or public devices.
• Use a secure device. Ensure that your device has up-to-date security measures such as antivirus software, firewalls, and the latest operating system updates.
• Be cautious with third-party integrations. Be careful when using third-party applications or services that integrate with ChatGPT. Ensure they have robust privacy measures in place.