We ask ChatGPT and similar chatbots questions all day long, without worrying about what happens to that data. What about ChatGPT and your privacy?
ChatGPT and your privacy: not so good
The major apps will probably all introduce their own version of ChatGPT in the coming year. Snapchat is already starting with My AI, while Twitter and Facebook are busy with their answer.
That is not surprising, because ChatGPT is an unprecedented success. The chatbot has quickly invaded everyday life. People in all kinds of professions use technology to make life easier for themselves.
Important documents or transcripts of meetings are also entered. Confidential conversations are placed in ChatGPT in this way to get a nice summary of it. It’s just important to be cautious about that, because privacy-wise, ChatGPT is a nightmare.
How AI learns
Such forms of artificial intelligence learn by feeding them a lot of data, after which connections are made. For example, ChatGPT has learned from a large pile of online texts. Not only for the knowledge in those articles, but mainly for understanding which word should follow in a logical sentence.
ChatGPT’s answers are based on these articles, but usually don’t resemble them word for word. Except when the chatbot spoons literal texts. If you ask for the first sentences of a book classic, you will just get those sentences – even if the text is copyrighted.
Many users see no problem with this, because access to information is more important to them. But what if the bot tells someone else what you told it in confidence?
What you tell these bots
Everything you type in the ChatGPT window is saved and used to further train the language model. If you enter a sensitive document, notes from a private conversation, or perhaps a piece of code to check, that information is now part of the ChatGPT database. It is possible that this information will later appear in the answers that others receive.
Anything you post online can be used to train AI. For example, Elon Musk is now working on his own version of ChatGPT and you can expect that chatbot to learn from what has been posted on Twitter. But while people are rightly cautious about what is posted on social media, we don’t yet have that restraint when it comes to chatbots.
Pretty funny that Elon Musk donated a billion dollars to OpenAI to be dissed by ChatGPT. pic.twitter.com/nfBHtJNt7L
— Alejandra Caraballo (@Esqueer_) December 6, 2022
These chatbots have proven to be very unpredictable and companies like Microsoft are now moving so fast with the technology that it has come at the expense of stability and security.
There is also no way to control what OpenAI stores about you, nor can you request that that data be deleted. It’s all swallowed up in the big picture and you don’t know if or when it will be spewed out again.
ChatGPT’s success shows that we like to throw our privacy overboard for a little bit of convenience. With the rise of the chatbot, it is important to be careful with that.