User conversations with the, da, were indexed and displayed in, compromising sensitive data and personal reports. The failure, revealed by the magazine Fast Companyinvolved at least 4,500 interactions links with, many containing confidential information such as mental health histories and trauma.
One of the examples cited by the report involves a user reporting experiences with posttraumatic stress disorder. Another described attempts to “psychological disprogramming” after supposed mental abuse.
The conversations, originally considered private, were accessible through a functionality called “Make this chat discoverable”(“ Make this chat accessible ”), inadvertently triggered by users.
Free tool
XP Simulator
Learn in 1 minute how much your money can yield
The vulnerability was admitted by Dane Stuckey, OpenAi’s head of security, in an X -post publication. According to him, the feature was a “short -term experiment” that allowed users to make public conversations to help others.
“We believe this feature has created many opportunities for people to accidentally share things they didn’t want,” Stuckey wrote.
Given the repercussion, OpenAi has announced the removal of the functionality and said it is working with search engines like Google to eliminate the already indexed links.
Continues after advertising
According to the company, the change will be implemented for all users this Friday (1st).
