A danger to privacy. On July 31, Openai announced a function that could make certain user conversations with chatgpt publicly accessible, through search engines such as Google.
For this, it required users to select a conversation that they wanted to share, then mark a box to share it with search engines. This is how almost 4,500 conversations unknown with ChatgPT appeared in Google’s search results, Fast Company magazine reported.
Sensitive conversations
However, this figure is likely to be really higher, Google does not indicate all conversations, as specified by the magazine. And according to the TechCrunch specialized site, it was even possible to find these exchanges filtering the search results, to include only the URLs of the domain name ‘https://chatgpt.com/share “.
If some were boring (finding recipe ideas, helping to renew their bathroom …), others were more personal and sensitive. In one of the conversations, a user, for example, described in detail his sex life and discomfort abroad, claiming to suffer a post -traumatic stress disorder and seek support. Another revealed to Chatgpt having survived a psychological program and trying to deprogram to mitigate the trauma suffered.
Worse, some conversations revealed personal information such as name or location. Given the risk posed by this function, OpenAi has decided to finish it.
This retropeding occurs a few days after Sam Altman, the OpenAi CEO, recalled that it was better to avoid trusting Chatgpt too much due to the absence of a legal framework in terms of confidentiality. A fault that could be problematic for the company in case of prosecution in court, because it could be legally obliged to reveal user conversations at the request of a court.
Source: BFM TV
