MIT Review Finds ChatGPT Surfaced 1 Real Address as Rivals Blocked Phone-Number Requests
Updated
Updated · Tech Times · May 14
MIT Review Finds ChatGPT Surfaced 1 Real Address as Rivals Blocked Phone-Number Requests
2 articles · Updated · Tech Times · May 14
MIT Technology Review found major AI chatbots handled requests for personal contact details unevenly, with ChatGPT returning an outdated but real phone number and address from an old government record.
Grok refused repeated requests for phone numbers, Claude cited privacy concerns, Perplexity blocked the attempts, and Gemini declined private numbers while sometimes identifying contact information already made public by a journalist.
The gap matters because conversational AI can pull data from websites, archives and public documents, making old or obscure personal information far easier to retrieve than through ordinary search.
Privacy advocates say that accessibility shift creates new risks, especially when chatbots surface inaccurate or unrelated numbers that could point users to innocent people.
Laws lag behind as AI exposes personal data. Who is liable when a chatbot reveals your private information?
AI can find your old address from forgotten files. Is it now impossible to truly erase your digital past?
Must we sacrifice our personal privacy for technological progress powered by our own data?
Millions of Personal Records at Stake: Privacy Gaps in ChatGPT and Generative AI Training Data
Overview
This report explores the privacy risks of AI chatbots like ChatGPT, focusing on how personal information such as phone numbers and addresses can sometimes be surfaced, even if outdated. While OpenAI and other developers have tried to prevent these leaks by designing chatbots to refuse sharing sensitive data, concerns remain about accidental exposure. The report also highlights that users sometimes give sensitive details to AI agents for convenience, as seen when a user provided Moltbot with credit card and login information. These examples show the ongoing challenges in protecting personal data as AI becomes more integrated into daily life.