OpenAI has removed a feature from ChatGPT that allowed shared chats to be discovered by search engines. The option in question only applied to chats that were shared by users.
According to a report by TechCrunch, some ChatGPT conversations were discoverable on Google, Bing, and other search services, when filters were used to include results from “https://ift.tt/wu32J8D". The report quotes various examples of data that could be visible on search engines. One person's LinkedIn profile was apparently easy to find based on the information in the chat. It is unclear whether other sensitive data was leaked.
While it does seem unintentional, it could have posed serious privacy issues. The issue was quickly patched by OpenAI, the company's Chief Information Security Officer (CISO), Dane Stuckey, announced on X that the feature was an experimental option that was designed to help people discover conversations they might find useful. It was an opt-in feature, i.e., a user had to click on "Share", and then click on a checkbox to allow the chat to be shared with search engines. Here is a screenshot of what the option for public link creation looked like.
User's names, custom instructions, and messages that were added after sharing remained private. There was also an option to control whether the shared link was to be discoverable.
Stuckey confirmed that the feature had been removed from ChatGPT. According to the statement, OpenAI decided to kill the option because it might lead to chats being shared accidentally. The company is also working to remove content that has been indexed by search engines.
Google has also released a statement that said neither it nor other search engines were in control of what pages are available publicly on the web, and it is up to publishers to decide whether a page is indexed by search engines.
Thank you for being a Ghacks reader. The post ChatGPT removes feature that let shared chats to be indexed by search engines appeared first on gHacks Technology News.
0 Commentaires