OpenAI is integrating a Health feature into ChatGPT, allowing users to connect their Apple Health data and medical records directly to the AI chatbot. This new functionality is now available to logged-in users aged 18 and older across all ChatGPT tiers (Free, Go, Plus, and Pro) on both web and iOS platforms.
The feature enables users to link Apple Health and, where applicable, medical records from U.S. hospital systems, One Medical, and Function Health. Once synchronized, ChatGPT can pull a range of personal health information, including medications, lab results, recent visit histories, sleep patterns, and activity logs, and weave them into any conversation within the app. This marks a significant shift from requiring users to access health-related information within a dedicated, separate section.
This redesign stems from OpenAI’s observations during an earlier, limited test of a similar feature. That version necessitated users navigating to a specific health area to receive responses grounded in their personal data. The company discovered that over 70% of health-related discussions among testers occurred organically within unrelated conversations, such as meal planning or queries about non-medical symptoms, rather than within the designated health space.
The current iteration addresses this by allowing ChatGPT to draw upon connected Health data contextually across any ongoing conversation, provided the user has explicitly granted permission. For example, a user discussing dinner plans might receive restaurant suggestions that take into account a registered dietary restriction. Similarly, someone planning weekend activities could be offered suggestions tailored to a recent injury noted in their synced records.
While a dedicated “Health” tab remains in the sidebar, its role has evolved into a centralized management hub. Here, users can connect and manage their accounts, review synchronized data and emerging trends, explore suggested prompts, and revisit past health-related conversations.
**Early User Experiences with ChatGPT’s Health Feature**
OpenAI has shared testimonials from early access users, which provide valuable insights into the tool’s practical application as a supportive resource rather than a diagnostic instrument.
Blake, a technical program manager, found significant value in transforming disparate medical history into an understandable and communicable format. He noted, “The most useful part has been turning scattered medical history into something I can actually understand and explain. I have multiple overlapping issues and ChatGPT helped connect those pieces into a clear timeline, explain the medical terms in plain English, and create summaries I could share with a physical therapist or trainer.” He further elaborated on the shift from disconnected records to identifying actionable patterns: “Instead of just seeing disconnected diagnoses, imaging results, and surgery notes, I could understand the bigger pattern. It made the information more usable and gave me better language to advocate for myself with providers and trainers.”
Reweti, a portfolio manager, highlighted the benefit of longitudinal analysis compared to standard searches or single doctor consultations: “What I want is to infer patterns that aren’t obvious and make connections I wouldn’t have made on my own. ChatGPT can access my existing labs because I’ve connected everything, and it can look over time—that’s the big advantage. It’s like having a research analyst. It allows me to be more proactive and own more of my health journey.”
However, not all experiences were entirely seamless. Shannon, a nurse, reported discovering an unexpected entry in her own medical chart via the tool, underscoring the sensitivity of surfacing clinical data through a chatbot. She emphasized the AI’s role as an enhancer, not a replacement, for healthcare professionals: “Using Health has actually reinforced something I’ve believed for a while: one of AI’s greatest strengths isn’t replacing healthcare professionals, but helping patients better understand and navigate their own health information. Discovering an unexpected chart entry through Health really highlighted that for me. It wasn’t AI creating a problem. It helped me identify something I can now appropriately follow up on with my healthcare providers.” This distinction between surfacing information for human follow-up and making clinical judgments is a critical line OpenAI must maintain as the product scales.
Other users, like Daniel, a consultant, focused on the practical benefits of data aggregation. He found the unified view derived from connecting multiple sources, such as Apple Health and MyChart, more valuable than isolated interactions: “It’s been great for coordinating labs and translating them into language I understand. Connecting Apple Health and MyChart makes the insights more grounded in what is happening across my life outside of just chat interactions.”
Kathleen, a small business owner, described a less critical but still impactful use case. The AI’s ability to track activity gaps in her synced data helped her re-establish a lost fitness routine: “I’ll say, ‘I need something to help me move today,’ and Health can see I haven’t worked out in the last seven days and suggest starting slowly—with a walk or some stretching. It’s helping me make things manageable and get back into it in a reasonable way.”
Carlton, an operations manager, previously spent time manually exporting and importing extensive spreadsheets from Apple Health for each chat. The new integration eliminates this manual effort: “Prior to Health, I was exporting massive spreadsheets from Apple Health and importing them into ChatGPT. Now, it feels a lot more streamlined. Being able to see years and years of my fitness journey in Health helped me see things in a different light—a bigger picture.”
OpenAI estimates that over 300 million health-related queries are processed by ChatGPT weekly, encompassing tasks from decoding lab results to preparing for medical appointments. If accurate, this figure positions ChatGPT as a dominant force in the digital health space, a position most similar platforms would take years and significant investment to achieve. The company’s clear positioning of Health as a supportive rather than diagnostic tool, coupled with explicit advice to consult healthcare providers for significant matters, is central to its strategy.
**OpenAI’s AI Model Advancements and Testing**
The efficacy of the Health feature is partly attributed to OpenAI’s newer AI models, GPT‑5.5 Instant for free users and GPT‑5.6 Sol for paid tiers. GPT‑5.5 Instant reportedly shows improved recognition of urgent care needs and enhanced explanation of uncertainty, performing comparably to OpenAI’s then-frontier models on challenging health assessments. GPT‑5.6 Sol is described as OpenAI’s most advanced health model to date, designed for complex reasoning across multiple data points, such as longitudinal lab trends.
To substantiate these claims, OpenAI collaborated with hundreds of physicians to develop health scenarios and scoring rubrics evaluating responses on accuracy, safety, communication, contextual understanding, completeness, and appropriate escalation pathways. OpenAI reports that GPT‑5.6 models consistently outperformed GPT‑5.5 on HealthBench Professional, an internally developed evaluation framework.
Internal comparative charts indicate GPT‑5.6 Sol achieved higher scores than GPT‑5.5 Instant and GPT‑4o across categories including accuracy, communication, completeness, and adherence to instructions, with physician-generated responses serving as a baseline. Furthermore, OpenAI states that physicians conducted specific testing of the live Health product prior to its release to assess real-world performance and safety with connected data, a crucial step beyond benchmark testing, although detailed methodologies and results from this testing have not yet been published.
**Health Data Management and Permissions**
OpenAI asserts that connected medical records and Apple Health data, along with any conversations incorporating this information, are excluded from foundation model training and targeted advertising, irrespective of a user’s general ChatGPT training settings. Conversations not involving health data adhere to the user’s pre-set training preferences.
Access to health data is permission-gated by default, with ChatGPT prompting users before utilizing connected health data to personalize a response. Users can opt for an “always allow” setting to disable these prompts, which can be managed within Settings > Plugins > Health and reversed at any time. Disconnecting a data source initiates deletion from OpenAI’s systems within 30 days. However, any health data previously surfaced in existing chat histories will remain accessible until manually deleted by the user.
Memory creation is also narrowly scoped and can be generated from health conversations but not directly from raw connected records or Apple Health data. Users seeking to prevent memory creation entirely can utilize Temporary Chat or disable memory features in their settings. OpenAI has also addressed an edge case where Health data could potentially be exposed through other connected plugins, such as sharing a training plan derived from Apple Health metrics with a fitness partner. The company states that additional safeguards are in place for such actions, and in sensitive situations, ChatGPT may request explicit user confirmation. OpenAI conducts red teaming exercises to identify vulnerabilities in these scenarios, though specific findings or failure rates have not been publicly disclosed.
**Addressing Accuracy at the Data’s Edge**
A significant practical challenge lies in the inherent variability of data quality. For instance, a medication entry might persist in a patient’s synced record even after they’ve discontinued its use. OpenAI uses such scenarios to underscore that synced data isn’t automatically current. Their guidance to users is direct: explicitly flag any data discrepancies to ChatGPT and always verify critical information against the original source rather than relying solely on the accuracy of the sync.
Data from wearables and fitness apps also presents limitations, as availability depends on what each third-party app shares through Apple Health. OpenAI acknowledges that some proprietary fitness app scores may not transfer at all.
The critical question extends beyond ChatGPT’s ability to accurately summarize a lab result in a controlled demonstration. It concerns the robustness of the permission model, data deletion protocols, and escalation logic when users’ synced records are outdated, and the AI is tasked with reasoning across conflicting datasets. While OpenAI’s physician testing addresses some aspects of this concern, it doesn’t fully bridge the gap between controlled evaluations and the complex reality of users managing multiple chronic conditions with incomplete data syncing from various sources.
For users interested in exploring this feature, it is currently accessible to eligible U.S. users via the sidebar Health menu.
Original article, Author: Samuel Thompson. If you wish to reprint this article, please indicate the source:https://aicnbc.com/24062.html
