Three rules to protect your privacy

From the talk

If you remember only one slide, remember this one. A KFF poll this year found that 13 percent of all adults, about 35 million Americans, have already uploaded their own medical information, like test results or doctor’s notes, into an AI tool. So let’s do it safely.

First, go to settings, then privacy, and turn off data sharing so your chats are not used to train the AI. Even then, your words pass through the company’s servers, which is why rule two matters. Remove names, birth dates, addresses, and insurance and record numbers before you upload anything.

Black them out with a marker or crop the screenshot. The medical content can stay, because that is what makes the answer useful. And rule three: treat every AI answer like advice from a smart but inexperienced colleague.

Read the slide as text

• About 35 million US adults have already uploaded their medical records into AI.
• Turn off data sharing
• Remove identifying details
• Verify before you act
• In settings, turn off the option that lets your chats train the AI.
• Names, birth dates, insurance and record numbers go. Medical content stays.
• AI is a draft, a suggestion, not a decision.
• “A 47 year old woman with pulmonary sarcoidosis” is fine. Your name and Medicare number are not.
• Sources: KFF, 2026; U.S. Census Bureau, 2025 (adult population estimate)