A significant privacy lapse has been discovered in Meta’s AI app, where users’ conversations are being exposed publicly without their knowledge or consent. The Meta AI app, integrated across Facebook, Instagram, and WhatsApp, has seen widespread use with 1 billion active monthly users. However, the app’s ‘share’ feature allows users to inadvertently post their conversations publicly, revealing sensitive information about medical, legal, and personal matters.
Privacy Concerns Raised
The issue came to light when it was found that users could access ‘shared’ conversations through the app’s Discover feed. A TechCrunch report highlighted instances of people seeking help with tax evasion and other delicate issues. The conversations revealed on the feed included a teacher sharing details about an upcoming arbitration for unjust job termination. The Meta AI responded with a supportive message, acknowledging the emotional weight of the situation.
Lack of Clear Guidance
Meta spokesperson Daniel Roberts stated that chats with Meta AI remain private unless users deliberately share them through a multistep process. However, critics argue that the company has not provided clear, in-app guidance at the time of sharing, leading to user confusion. When questioned about these mishaps, the Meta AI chatbot acknowledged that users might unintentionally share sensitive information due to misunderstandings about platform defaults.
Protecting Your Conversations
To maintain privacy when using Meta AI, users can take several steps:
- Disable Public Sharing: In the Meta AI app, go to Data & Privacy under App settings, then select Manage your information and set ‘Make all your prompts visible to only you’.
- Avoid the Share Button: Refrain from using the Share button unless you intend to make your conversation public.
- Limit Data Use for AI Training: On WhatsApp, Facebook, and Instagram, navigate to Settings & Privacy > Privacy Center, then scroll to Privacy Topics and select AI at Meta. Submit an objection request to limit Meta’s use of your data for AI training.
- Delete AI Conversations: Use commands like /reset-ai in Messenger, Instagram, or WhatsApp to delete AI conversation data.
Meta’s handling of user data has been criticized for lacking transparency and clear guidance. As the company continues to expand its AI capabilities, concerns about privacy and data security remain paramount. Users must be aware of the potential risks and take proactive steps to protect their conversations.
