Contract workers for Meta have reported regular exposure to personal information, such as full names and phone numbers, while reviewing user interactions with AI chatbots. This process is part of the training effort where contractors rate and assess the AI's responses. An estimated half of the chats reviewed included identifiable information, and some users had even shared selfies. Despite Meta's privacy policy discouraging such disclosures, many users inadvertently provided personal data during conversations, raising significant privacy concerns about the handling of sensitive information.
Contract workers hired by Meta reported regularly encountering personal information, including full names and phone numbers, in conversations between users and the AI chatbot.
One contractor estimated that personally identifiable information appeared in over half of the thousands of chats reviewed each week, which included selfies and detailed personal data.
Meta's privacy policy warns users not to disclose personal information in chats; however, users often revealed sensitive details during conversations with the AI.
Contractors indicated that the inclusion of unredacted personal data was more prevalent in Meta projects compared to similar projects for other Big Tech clients.
Collection
[
|
...
]