Is ChatGPT Putting Your Data Privacy at Risk?

Over the last few months since its release, generative AI ChatGPT has created a lot of excitement for its potential to revolutionize how we communicate with computers — and maybe even each other.

This AI uses language processing that mimics natural conversation convincingly and in real-time, giving users answers and access to information in an advanced manner.

ChatGPT is more advanced than any conversational AI the public has had access to up till this point, and it’s little wonder that it’s making waves.

But along with the novelty and the potential for automation and other use cases comes a slew of security and privacy concerns.

How Does ChatGPT Work?

ChatGPT takes a user’s input, such as a written prompt, and creates an output of text information; it’s referred to as text-to-text, or as some users prefer, text-to-essay.

ChatGPT outputs can be anything from a memo, to a scientific paper, to a blog article, to a draft of a legal document. (If you ask, it can even crank out poetry on space travel in the style of a Shakespearean sonnet.)

Receiving an output is as simple as typing in a prompt, then copying and pasting the results.

Behind the scenes, ChatGPT is comparing and compiling information in smaller-than-word chunks called tokens. It breaks down the input text and then recompiles these tokens into what it deems the most likely string of tokens to satisfy your input.

However, before you start diving into ChatGPT or other similar tools, be aware that any information you supply could become part of the collective of information ChatGPT can and will draw on for further comparison, analysis, and compilation.

Security and Privacy Problems

Because ChatGPT is so easy to use, some people have been all too happy to share personal or proprietary information without realizing the ramifications.

First, if this personal and proprietary information is shared and becomes part of the collective, there is no simple delete process to remove it. You don’t have control over what happens to that data once you give it to the AI.

Practically, this means that while ChatGPT could, in theory, provide useful text in just about any professional context, users must exercise caution — especially in industries that use protected or proprietary information.

It’s possible that using ChatGPT improperly could violate privacy regulations or divulge business secrets. Job loss, fines, and possibly even legal action could result from improper use.

Should Businesses Use It?

Businesses that use ChatGPT must weigh the information they share with the AI and take care not to share anything that needs to remain private or proprietary.

There is a non-trivial chance that this information could eventually become widely accessible to the AI tool, even showing up in future outputs. And there’s always the possibility that information could leak as part of a breach.

Many current and potential applications of this tool could function while working around revealing such sensitive information.

Users should give careful consideration to what terms and ideas to remove or alter before uploading inputs, whether the output is something they’d like available to others once created, and whether it is possible with the particular AI chatbot they have chosen to set up only a local instance of the tool or if it remained on a public cloud.

Other Dark Sides to Conversational AI

As intriguing and powerful as conversational AI like ChatGPT is, be aware of certain limitations or dark sides.

First, a consumer threat: bad actors could easily use an AI like ChatGPT to create natural-sounding phishing emails and web page content.

Second, be aware that ChatGPT cannot reliably distinguish between truth and falsehood in ways that would be obvious to the average person. It has been shown to generate completely made-up references (citing a nonexistent article by a made-up name but in a real journal).

It can confidently assert incorrect information right next to correct. It can only draw from the information it can access and may fill in the gaps with plausible-sounding content.

Ultimately, remember that ChatGPT is not a person nor a super-intelligent computer. It’s an AI trained to generate, one token at a time, the text it decides is the most likely to match your input. Sometimes that text will be amazing.

Other times it’s flat-out wrong. And use extreme caution before feeding it any sensitive, protected, or proprietary information.