Highlights
Meta AI App Faces Scrutiny Over Privacy Concerns
Meta AI app is encountering increasing criticism as users have been found to inadvertently disclose sensitive and often personal conversations to the public. Launched in April, the independent Meta AI application features a social feed-style “Discover” tab that highlights interactions between users and the chatbot. These exchanges cover various topics, including travel recommendations and recipes, but also extend to concerning admissions and confidential inquiries.
Unintended Sharing of Personal Data
While conversations are not publicly shared by default, the application incorporates a noticeable “share” button. This feature enables users to publish their discussions on a public feed. However, critics contend that the interface does not adequately clarify when content is being made public. This has led to a surge of posts exposing names, locations, contact numbers, legal issues, and health conditions.
The public feed displays numerous instances of users seeking guidance on matters such as tax evasion, court character references with full names, and information regarding potential corporate fraud. Additionally, there are shared medical concerns related to skin rashes, bowel health, and post-operative recovery questions, many of which can be traced back to public Instagram profiles associated with their Meta accounts.
Meta’s Response to Privacy Concerns
In a blog post, Meta asserted that “nothing is shared to your feed unless you choose to post it.” The company also highlighted that the AI experience becomes increasingly personalised when using data from Meta’s other platforms, including Instagram. This emphasis on personalisation raises questions about user awareness and data visibility.
Calls for Improved User Protection
Despite the fact that the application has been downloaded around 6.5 million times, reports indicate a growing lack of confidence regarding the exposure of identifiable information. Experts believe Meta must reassess its strategy concerning AI interactions and reconsider assumptions about how users prefer their conversations to be shared.
