Meta’s new AI app is under fire after users unknowingly shared private chats, including legal queries, personal data and audio clips, on the public web.

The issue lies with a “share” button that appears after each chatbot response. Users can post content without realising it’s publicly visible, especially if logged in via a public Instagram account. Security expert Rachel Tobac called it a “privacy nightmare” after spotting names, addresses and court-related questions shared online.

Some posts appear jokey or attention-seeking, but many involve sensitive or reputationally risky content. One user asked about a rash, another discussed tax evasion, and several uploaded CVs and legal references, seemingly unaware they were going public.

Launched on 29 April, the app has already hit 6.5 million downloads. However, experts say Meta should have anticipated the risks of blending private AI queries with social sharing.

Businesses should avoid using AI tools through personal logins and steer clear of sharing anything sensitive unless privacy settings are crystal clear.