Openai feature allows Chatgpt users to share their online conversations and discussions. As a result, the Google search engine recovered them and nearly 100,000 conversations could be accessible to everyone.
This is not a case of hacking or other, but a lack of rigor in the intentions of Openai. Concretely, tons of sensitive and confidential information found themselves in the open air on the Internet and accessible to anyone who knows how to seek.
From confidential contracts to marital problems
This story published by 404 Media reveals that the set of data includes a wide range of sensitive information, other harmless, alleged texts of confidentiality agreements, discussions on confidential contracts, people trying to use Chatgpt to analyze their relationship problems, and many people asking Chatgpt to write LinkedIn publications.
This news is following an article published on July 30 by Fast Company, which reported that “thousands” of shared chatgpt conversations appeared in Google search results. Since then, people have searched some of the conversations indexed by Google. The data set of around 100,000 conversations gives a better idea of the extent of the problem and highlights some of the risk for privacy linked to the use of AI tool sharing features.
Quickly deleted feature
Dane Stuckey, business systems security manager for the company, told 404 Media that he had removed the Chatgpt functionality which allowed users to make their conversations accessible to search engines, such as Google. It was a short -lived experience to help users discover useful conversations.
As Fast Company pointed out in his report, it is not certain that Chatgpt users understood that they were making their conversations to everyone’s sight.