ChatGPT Health: What It Means for Your Data

TL;DR: OpenAI has launched "Health in ChatGPT," a feature that lets users store, share, and discuss health data — including records from Apple Health and Google Fit — directly in the ChatGPT app. While this pushes AI deeper into consumer health, it raises real questions about data privacy and HIPAA compliance. For developers building health-related tools, open-model APIs like those on TokShop offer a privacy-focused alternative with transparent per-token pricing.

What Is Health in ChatGPT?

OpenAI's "Health in ChatGPT" is a new feature that turns the chatbot into a personal health hub. It allows users to connect health data sources like Apple Health, Google Fit, and Samsung Health, and then ask ChatGPT questions about their sleep, activity, and other metrics. The feature also supports manual entry of health information, such as medications or symptoms, and can generate summaries for doctor visits.

The rollout is gradual — OpenAI says it's "widely available" but still rolling out to free and Plus users over time. The feature is built on OpenAI's existing privacy architecture, meaning conversations are stored and can be reviewed by the company unless users opt out of training. This is the first time OpenAI has moved this deeply into the consumer health space, and it signals a broader push to make ChatGPT a daily health companion.

How Does OpenAI Handle Patient Health Records?

OpenAI has stated that Health in ChatGPT is not HIPAA-compliant for covered entities, and it does not offer business associate agreements (BAAs) for this consumer feature. That means if you're a healthcare provider, you cannot legally use ChatGPT to store or process patient records under HIPAA rules. For individual users, OpenAI says health data is treated like any other conversation data — it can be used to improve models unless the user opts out.

The company has added some safeguards: a new "Health" tab in the app, a dedicated privacy policy section for health data, and an option to export or delete health data. However, OpenAI's terms still allow it to share data with service providers and, in some cases, law enforcement. There's also no end-to-end encryption for health conversations — data is encrypted in transit and at rest, but OpenAI employees can access it for safety and abuse monitoring.

What Are the Privacy Risks Developers Should Know?

The biggest risk is that consumer health data in ChatGPT is not protected by HIPAA, and there's no clear path to making it so. Even if a developer builds a wrapper around ChatGPT for health use cases, the underlying data handling remains under OpenAI's consumer terms. That creates a compliance gap for any app that touches patient data.

Another concern is data retention. OpenAI stores conversations unless users manually delete them, and even then, deletion may not remove all copies from backups. For health data, this is a significant issue — a user might share sensitive information in a moment of need, only to have it linger in OpenAI's systems indefinitely. Developers should also note that OpenAI's API (which powers custom apps) has different data policies than the consumer ChatGPT app, but the API is still not HIPAA-compliant without a BAA.

Can Developers Build Health Apps Without ChatGPT?

Yes — and many are choosing open-model APIs to keep health data out of proprietary systems. TokShop offers OpenAI-compatible endpoints for models like DeepSeek V3.2, GLM 4.6, and Kimi K2, which you can use to build health-related features without tying your app to OpenAI's consumer data policies.

Here's a simple example of how you'd call a TokShop model for a health-related query using Python:

from openai import OpenAI

client = OpenAI(
    base_url="https://tokshop.xyz/v1",
    api_key="sk-tok-..."  # your TokShop key
)

response = client.chat.completions.create(
    model="deepseek-v3.2",
    messages=[
        {"role": "system", "content": "You are a health assistant. Provide general wellness information only."},
        {"role": "user", "content": "What are common signs of dehydration?"}
    ]
)

print(response.choices[0].message.content)

With TokShop's pay-as-you-go pricing, you only pay for what you use — DeepSeek V3.2 costs $0.42 per million input tokens and $0.63 per million output tokens. That's a fraction of what you'd pay for comparable proprietary models, and you maintain full control over where the data goes.

What Should Developers Consider Before Choosing a Health AI API?

First, decide whether your use case involves protected health information (PHI). If it does, you need a HIPAA-compliant solution — which means a BAA from your API provider. Neither OpenAI's consumer ChatGPT nor most open-model APIs offer this by default, so you'd need to process data on your own infrastructure or use a specialized healthcare API.

Second, consider data residency. If you're serving users in the EU or other regions with strict data laws, you need to know where your API provider stores data. TokShop doesn't advertise specific data residency guarantees, so for regulated health data, you should contact the provider directly.

Third, look at the model's ability to handle medical context. Models like GLM 4.6 (with a 200,000-token context) and Qwen3 Coder (262,144 tokens) can process long medical documents, but they're not trained specifically for clinical accuracy. Always include a disclaimer that your app provides informational content, not medical advice. For a full breakdown of pricing and context windows, check the TokShop pricing page.

FAQ

Is Health in ChatGPT HIPAA compliant?

No. OpenAI's consumer ChatGPT feature is not HIPAA-compliant and does not offer business associate agreements for health data. Covered entities cannot use it to store or process patient records.

Can I use TokShop models for health-related apps?

Yes, TokShop's OpenAI-compatible API can be used for health-related features like wellness tips or symptom checkers. However, you're responsible for ensuring your app complies with applicable regulations, and you should not use it for PHI without a BAA.

How much does it cost to build a health chatbot with TokShop?

It depends on the model and usage. For example, DeepSeek V3.2 costs $0.42 per million input tokens, making it affordable for high-volume, low-complexity health queries. Check the TokShop docs for integration details and current pricing.