Meta pulls back on AI chatbot prompts after personal-data backlash

Meta is reworking the conversation prompts generated by its Meta AI assistant after a widely shared social media video showed the chatbot suggesting questions about a user’s children, according to a report from The Verge.

The incident began when Instagram user Kalie Robins posted a clip describing how the AI feature surfaced an intrusive suggestion after she cross-posted a video to Facebook. She said the assistant displayed a prompt reading “Who is the child passenger?” beneath footage featuring her and her child, leaving the question of why the system was drawing attention to the child’s identity at all.

In a statement to The Verge, Meta spokesperson Dina El-Kassaby acknowledged the company “missed the mark,” adding that the feature “never should have prompted the individual with questions like that.” The company said it is making changes to how such suggested prompts are generated.

The episode highlights a growing tension in consumer AI products: assistants that draw on personal context — photos, captions, relationships, and activity across linked accounts — can produce responses that feel invasive even when no rule is technically broken. Meta’s AI features span Instagram, Facebook, WhatsApp, and Messenger, meaning a misfiring suggestion can travel across platforms where users share family content, including material involving minors.

The backlash also lands at a sensitive moment for the broader tech sector, where companies across industries — from social platforms to Block, Inc., the payments and financial software company whose shares closed at $79.40, down 1.0% from a previous close of $80.20, valuing it at roughly $47.7 billion — are racing to weave generative AI into consumer-facing products while navigating privacy expectations and regulatory scrutiny.

Meta has not detailed the specific technical changes it plans to make to its prompt suggestions, nor how it will filter AI outputs that reference children or other sensitive personal details. The company’s response, delivered within days of the video going viral, suggests the prompt was not behaving as designed, but it remains unclear what guardrails existed before the change.

Privacy advocates have long warned that AI assistants with access to personal context blur the line between helpfulness and surveillance, particularly for users who do not realize how much information the systems are inferring. The Robins video added a high-profile example to that debate.

What to watch

• Details from Meta on how it will restrict AI-generated prompts involving children or family members.
• Whether similar prompt-suggestion complaints surface on other platforms where Meta AI operates, including WhatsApp and Messenger.
• Any regulatory or policy responses concerning AI systems processing content involving minors.
• Upcoming Meta earnings commentary on AI feature rollouts and user-safety measures.

Source: original release