Experts Warn 50% of Americans Against Sharing Sensitive Data With AI Chatbots
Updated
Updated · HuffPost · Sep 12
Experts Warn 50% of Americans Against Sharing Sensitive Data With AI Chatbots
3 articles · Updated · HuffPost · Sep 12
Summary
Half of Americans now use AI chatbots, but experts say users should stop treating them like private spaces for personal, medical, financial or work-related disclosures.
A Stanford review of six major U.S. chatbot developers found all appear to use conversations to train models by default, with some retaining data indefinitely.
The highest-risk material includes personally identifiable information, intimate details, health records, confidential business data, source code, banking details and tax documents—information that could fuel fraud, breaches or legal exposure.
Users who already shared sensitive data are urged to delete chat histories, tighten privacy settings, opt out of training where possible, and rewrite prompts with pseudonyms or generalized descriptions.
Researchers say stronger guardrails should not rest on users alone, calling for federal privacy rules, affirmative opt-ins for model training and default filtering of personal information.