Apple is preparing a major update to its Health app that will introduce more artificial intelligence features. Still, the move is already raising concerns among users familiar with Google Health’s troubled rollout. The new experience is expected to use AI to analyse health and wellbeing information and provide users with more personalised guidance.
Apple highlighted the changes during its recent iPhone event, where the company outlined broader plans for AI-powered health features. While the technology could make it easier for people to understand their health data, Google’s experience with a similar approach has shown the potential risks of relying on AI for sensitive wellbeing advice.
Google Health’s AI has faced criticism from Fitbit users
Google began replacing the Fitbit companion app with its redesigned Google Health experience in May 2026. Fitbit users criticised the transition, citing missing features, interface changes, and a less intuitive experience than the previous Fitbit app. Some users described the new application as “unbelievably bad”.
The criticism intensified when Google introduced an AI assistant designed to examine users’ health information and offer recommendations on exercise, fitness, and general wellbeing. Rather than simply presenting existing information, the assistant was intended to interpret data and provide more personalised guidance.
However, user reports suggested the AI could produce unreliable or confusing responses. Complaints included what users described as “unhinged” fitness advice and “nonstop lies”, while some reported that the system generated fictional information relating to food tracking and other health metrics. Such errors can be particularly frustrating when users expect an app to provide dependable information based on data collected from their wearable devices.
The problems highlight a broader issue with applying generative AI to health applications. An AI system can produce a convincing response even when the underlying information is incorrect. In a typical productivity or entertainment app, an inaccurate answer may be inconvenient, but health-related recommendations can have more serious consequences.
That makes Google Health an important example for Apple as it develops its own AI-powered health experience. The company will need to ensure that any recommendations generated by its system are based on reliable information and presented with appropriate limitations.
Apple’s approach could be different
Apple has generally taken a more cautious approach to introducing artificial intelligence into its products, although its wider Apple Intelligence strategy has faced its own difficulties. The company has previously pursued its own AI capabilities but has also turned to Google for assistance as it works to improve Siri and other AI-powered functions.
That relationship raises questions about how much of Google’s underlying technology could eventually influence Apple’s health features. However, using technology associated with Google does not necessarily mean Apple Health will behave the same way as Google Health. Apple could apply its own software, safeguards and health-specific systems around any external AI models it uses.
Apple also says its approach to health technology involves extensive testing and specialist input. The company states that “all of our health innovations are subject to rigorous scientific validation and are developed with clinical experts from start to finish”. That commitment will matter as AI becomes more involved in interpreting personal health information.
The distinction between generating general information and making health recommendations is significant. Apple Health already collects and displays data from compatible devices and services, but an AI system that can interpret that information adds another layer of complexity. The system must not only understand the data but also avoid confidently drawing conclusions it cannot support.
Privacy is another major consideration. Health information is among the most sensitive types of personal data, and Apple has repeatedly highlighted its approach to encryption, security and on-device processing when discussing health-related features. The company will face considerable expectations from users to maintain those protections as AI becomes more deeply integrated into Health.
Apple Health’s AI will face a crucial test
Apple therefore has an opportunity to learn from the problems Google Health encountered before its own redesign reaches users. The criticism surrounding Google’s AI assistant demonstrates how quickly confidence can disappear when a health application produces recommendations that appear inaccurate or nonsensical.
For Apple, avoiding similar problems will require more than simply connecting an AI model to existing health data. The company will need to make sure its system can distinguish reliable information from uncertainty and communicate those limitations clearly. It will also need safeguards to reduce the possibility of misleading recommendations being presented as authoritative health guidance.
There is no certainty that Apple Health will repeat Google Health’s mistakes. The companies may use different implementations, data sources and safety systems, even where some underlying AI technology overlaps. Apple’s existing focus on health research, privacy, and clinical involvement could also add layers of protection.
However, the comparison is difficult to ignore. Google Health’s rollout has shown that adding an AI assistant to a health platform can create new problems alongside its potential benefits. Apple will now have to demonstrate that its approach can provide useful personalisation without sacrificing accuracy or user trust.
The new Apple Health experience is not expected to arrive immediately. Apple has said the updated features will become available “later this year”, meaning users will have to wait before they can assess how the system performs in practice. Until then, Fitbit users’ experience warns what can happen when AI is given too much responsibility in a health app without sufficient reliability.
