AIToday
Large Language ModelsAI WeeklyPublished: Nov 21, 2026, 06:00 JST

ChatGPT Health hits Mac, Android, Dots; health chats now train models

Send on LINE
ChatGPT Health hits Mac, Android, Dots; health chats now train models

3 Key Points

  1. What happened

    OpenAI extended ChatGPT Health from iPhone, iPad and web to the Mac desktop app, Android via Health Connect, ChatGPT Sites and the Dots proactive assistant on November 19; the feature stays US-only.

  2. Why it matters

    From November 19, chats that draw on connected health data can be used to train OpenAI's models unless the user first switches off the model-improvement setting in Data controls, reversing a prior guarantee that such data stayed out of training.

WHO IT HITSUS-based ChatGPT users who connect Apple Health or medical records to the assistant now bear the burden of checking a data-control setting if they want their health conversations excluded from model training. The expansion to Android and Dots adds new surfaces where health context can enter ChatGPT.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Context & Analysis

The expansion itself follows a pattern OpenAI has used before: ChatGPT Finances also started on mobile and web before reaching the Mac app, so the desktop rollout of Health is consistent with how the company stretches a feature across surfaces. What is different this time is the policy shift that landed on the same day. According to Pasquale Pillitteri's write-up, OpenAI's earlier commitment was that connected health data and the conversations using it stayed out of training regardless of the model-training setting. From November 19, that setting decides whether such conversations can enter training, and content derived from the data may be used if the setting is on. The change was announced with six days' notice, and Pillitteri frames it as moving a guarantee behind a toggle, which shifts consent from automatic to something the user owes. The simultaneous arrival on Android through Health Connect and on the Dots proactive assistant adds two more places where health data can enter ChatGPT's context, one of them ambient and not prompted by the user. The feature remains US-only, so the practical effect is concentrated on US users who connect health records or Apple Health.

FAQ
How do I keep my ChatGPT Health conversations out of model training?
You need to go to Settings > Data Controls and disable 'Improve the model for everyone'. From November 19, that setting decides whether chats drawing on connected health data can enter training.
Which platforms now support ChatGPT Health?
It is available on iPhone, iPad, web, the Mac desktop app, Android via Health Connect, ChatGPT Sites and the Dots proactive-assistant feature. The rollout is US-only.
What changed about OpenAI's prior privacy guarantee for health data?
OpenAI had guaranteed that connected health data and the chats using it stayed out of training regardless of the model-training setting. Effective November 19, that protection now depends on the user turning off the model-improvement setting.
Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleDragos's HD Moore finds 3 AI providers breached in 2M-prompt test