How AI chatbots surface personal data from data brokers — and how to shrink what they can reveal about you.
In late 2025, reporters asked Grok’s free web version for the home addresses of 33 ordinary, non-famous people. It returned correct addresses for 10 of them — often with phone numbers and family members attached. No hacking. Just a question.
Web-connected AI models don’t “know” your address — they retrieve it. People-search sites and data brokers publish and resell your home address, phone, relatives, and past addresses. A model cross-references those sources in seconds, doing in one prompt what used to take a determined stalker weeks.
Some models refuse these requests on privacy grounds. Others answer. Your safety shouldn’t depend on which app the wrong person opens.
The reason people pay for this is simple: broker removal decays. Do it once and you’re clear for a season; ignore it and the data flows back. Ongoing removal + monitoring is what keeps your exposure shrinking instead of creeping back.
Sources: Futurism (Grok address findings, 2025); California Privacy Protection Agency DROP platform, 2026.
Our scan includes a privacy probe — see if the models expose your location. Free, ~60 seconds.
Scan my name →