“What county is it [sic] Public information from users on Meta’s AI platform says: “I need details, I’m 66 years old, single. I’m from Iowa and if I can find a young woman, I can be willing to move to a new country.” The chatbot enthusiastically replied: “You are looking for a new starting point and love in a new place. It’s exciting!” Before proposing “Mediterranean countries like Spain or Italy, or even Eastern Europe.”
This is just one of many seemingly personal conversations that can be viewed publicly on Meta AI, a chatbot platform that can be used as a social feed and launched in April. In the metaAI application, the Discover tab shows a timeline for others to interact with the chatbot; the short scroll on the metaAI website is an extensive collage. While some highlighted queries and answers are harmless (Trip itineraries, recipe suggestions), others reveal locations, phone numbers and other sensitive information, all closely related to usernames and profile photos.
Calli Schroeder, a senior consultant at the Electronic Privacy Information Center, said in an interview with Wired that she has seen people “sharing medical information, mental health information, home addresses, and even directly related to pending court cases.”
“All of this is incredible, both because I think it points out how people misunderstand the work of these chatbots or their purpose, and how privacy works with these structures,” Schroeder said.
It is not clear whether users of the app realize that their conversations with Meta’s AI are public or that users are dragging the platform after news outlets start reporting. By default, conversations are not public. Users must choose to share them.
There is insufficient conversation between users and Meta’s AI chatbot, which seems to be private. One user asked the AI chatbot to provide a format to terminate a renter’s lease, while another asked it to provide an academic warning notice that provides personal details, including the school name. Another asked the sister for the responsibility for potential tax fraud in a particular city in a particular city using accounts related to Instagram profiles showing first and last names. Others asked it to file a role statement with the court, which also provides numerous personally identifiable information about the alleged offender and the user himself.
There are many examples of medical problems, including people struggling in defecation, seeking help from the hive, and asking about rashes on the inner thighs. One user told Meta AI about their neck surgery and included their age and career in the prompts. Many, but not all, accounts seem to be related to personal Instagram profiles.
Meta spokesman Daniel Roberts wrote in an emailed statement that unless users share them on discovery feeds through a multi-step process, users’ chats with Meta AI are private. The company did not answer questions about mitigation issues with sharing personally identifiable information on meta-AI platforms.
