TECHnicalBeep

Find out how OpenAI addressed security issues in their ChatGPT macOS app by encrypting conversations that are saved. Understand the objective of data protection in Artificial Intelligence technologies.

Background:

Last month a vulnerability issue was discovered in OpenAI’s ChatGPT macOS app. Initially, the conversations between patients and physicians were saved in simple text messages that everyone could access. This causes much concern as to the privacy of the users and the security of their information.

The Security Flaw Unveiled:

Pedro José Pereira Vieito explained how it was possible to obtain the local CHatGPT conversations from the users’ computers. Ideally, with the help of a simple app, Pereira Vieito could have followed these conversations as soon as they took place and that shows how ill-advised and insecure such conversations are. This means that the user’s information was vulnerable with possible malicious parties getting access to the users’ devices.

OpenAI’s Swift Response:

In this regard, it was after The Verge had brought the matter to their notice that OpenAI took the necessary steps. We must admit that they issued an update that encrypts all the locally created ChatGPT discussions. OpenAI representative Taya Christianson stressed the company’s focus on security while also keeping ease of use in mind.

Related Content: ChatGPT – Understanding the Next Generation of Conversational AI

The Technical Insight:

Vieito of Chulalongkorn University, the person who discovered the weakness, pointed out that OpenAI failed to use app sandbox protection which, if implemented, would have blocked unauthorized access. This decision enabled his app which was intended to showcase the frailty to quickly load and present the ChatGPT interactions.

Looking Ahead:

Although OpenAI does monitor conversations for safety and training of the model, which was the case in this incident, it shows that security should be improved. For the time being, customers assume that the information to be processed must be protected not only from the AI supplier but also from various dangers beyond this supplier.

Conclusion:

Altogether, it can be stated that OpenAI responds actively to the given initial security issues, specifically by encrypting the conversations with ChatGPT. However, as AI moves forward it is still important to remember the security of data. Some changes are as follows, the users can continue using the updated ChatGPT macOS apps, with the guarantee that their conversations are safe.

Image Credit: OpenAI

author

Leave a Reply

Your email address will not be published. Required fields are marked *