OpenAI has made the availability of ChatGPT Health, its health and wellness focused assistant, public for all adults residing in the United States. This launch occurred this week and is accessible in both free and paid versions, on both the web interface and iOS devices.
This move comes one day after a lawsuit filed by a Florida pastor. He alleged that he received advice from the artificial intelligence that could be fatal, as he was advised not to seek specialized medical assistance.
The tool's functionality is based on cross-analysis of clinical records and metrics obtained from healthy living applications, utilizing the computational model GPT 5.6-Luna. The system registered an impressive volume of 300 million requests every seven days.
Due to the high volume of use, the system has been updated to allow diagnoses and dietary habits to be consolidated in any conversation section within the ecosystem, expanding its applicability beyond the restricted dashboard previously used only by volunteers in tests.
The ecosystem's infrastructure allows for the unification of databases from large hospital networks, such as Epic and Oracle Health, in addition to integrating services like One Medical, Apple Health, Function, and MyFitnessPal.
The developer emphasized that any private data entered into the dashboard is not used to train its algorithms. Furthermore, she highlighted the continuous participation of medical professionals in refining the responses provided.
Although the GPT 5.6-Luna model demonstrated superior performance compared to the GPT 5.5 version metrics in the HealthBench test, the corporation reiterated in its terms of use that the mechanism does not have the capacity to issue diagnoses or prescribe treatments for diseases. The company's advisory informed the press that it strongly recommends confirming any guidance with human specialists.
Recent academic research indicates inconsistencies in virtual robots when offering clinical diagnoses. However, this scenario of uncertainty has not prevented the advancement of large technology companies, as competitors such as Google and Anthropic continue to invest in developing similar tools aimed at human health.
On Thursday afternoon, the 23rd, ChatGPT began to exhibit various operational problems. The Downdetector website registered a significant increase in complaints regarding the platform, and the official OpenAI status page confirmed the occurrence of system failures.
On Downdetector, the number of complaints reached over 2,700 reports around 3:46 PM Brasília time. The service's official status page reported that 'some users may experience high levels of errors in ChatGPT,' adding that OpenAI has already implemented fixes and is monitoring the system.
A test conducted by Olhar Digital around 3:55 PM showed that, although ChatGPT was slow to process and generate responses, access was successful.
As usual, many individuals from various parts of the globe expressed their dissatisfaction with the situation on the social network X. Some users even associated this Thursday's problem with an alleged hacking of ChatGPT.
