Alexander Culafi , Senior News Writer , Dark Reading January 19, 2026 6 Min Read Source: Panther Media Global via Alamy Stock Photo The recent announcement of LLM health chatbot product ChatGPT Health suggests a world where health advice will be at the consumer's fingertips more than ever before, but with the product also comes a wide range of safety and data security concerns. On Jan. 7, OpenAI announced ChatGPT Health , described by the LLM firm as "a dedicated experience that securely brings your health information and ChatGPT’s intelligence together, to help you feel more informed, prepared, and confident navigating your health." The company sites health concerns as one of the most common ways individuals interact with ChatGPT, and Health intends to be a more secure dedicated experience. It comes with "additional, layered protections designed specifically for health," OpenAI said, such as "purpose-built encryption and isolation to keep health conversations protected and compartmentalized." OpenAI also stresses that ChatGPT Health is designed to support and not replace medical care, is not intended for diagnosis or treatment, and that the data shared with the product is not used to train its foundation models. In a world where hundreds of millions of people are already using AI chatbots for health inquires — regardless of whether such a thing is advisable — it may in theory be welcome to have a siloed-off product that protects the most sensitive personally identifiable information an individual has. And yet, it must be stressed that ChatGPT Health enables the user to connect one's own medical records with it, and those records can be shared with third parties like Apple Health and other wellness apps, if the user opts in. The user would be entrusting a private company with his or her most sensitive information in order to solicit medical advice. Even in a perfect world with flawless encryption, no prompt injection attacks, an excellent data security track-record , and no severe safety concerns , such a move is arguably risky. And we do not live in a perfect world. ChatGPT Health's Security Features ChatGPT Health's announcement touches on how conversations and files in ChatGPT as a whole are "encrypted by default at rest and in transit" and that there are some data controls such as multifactor authentication, but the specifics on how exactly health data will be protected on a technical and regulatory level was not clear. However, the announcement specifies that OpenAI partners with network health data firm b.well to enable access to medical records. An OpenAI spokesperson tells Dark Reading that b.well pulls patient records via Individual Access Services requests, and that the company is built on Fast Healthcare Interoperability Resources (FHIR) standard APIs as well as trusted healthcare exchange frameworks. Additionally, the spokesperson says the b.well network maintains strict identity verification, consent management, data security, and auditability controls. As for third party connections such as wellness apps and medical records, the spokesperson explains that all apps in Health must meet OpenAI privacy and security requirements, collect the minimum data needed, and undergo additional review. Users can also disconnect an app at any time, at which point the app loses access. Franco Giandana Gigena, policy analyst at digital rights nonprofit Access Now, tells Dark Reading that "once data has been shared, it is almost impossible to completely and entirely delete it, and it is often indicated that this means that data control is lost." Even if a user can choose to disconnect from a third-party app, that only prevents further data sharing. Ultimately, users should be careful before trusting any product with their health data. "We would like to underline that in health-related contexts, sometimes consent may not always be sufficient to protect data in complex webs of data sharing," Giandana Gigena says. ChatGPT Health's Troubling Question Marks While many security tentpoles remain in place, healthcare data must be held to the highest possible standard. It does not appear that ChatGPT Health conversations are end-to-end encrypted (Dark Reading asked and the spokesperson did not directly say; "in transit and at rest encryption," as the announcement mentions, is not end-to-end). Regulatory consumer protections are also unclear. Dark Reading asked OpenAI whether ChatGPT Health had to adhere to any HIPAA or regulatory protections for the consumer beyond OpenAI's own policies, and the spokesperson mentioned the coinciding announcement of OpenAI for Healthcare , which is OpenAI's product for healthcare organizations which do need to meet HIPAA requirements. But compared to the educationally focused consumer product ChatGPT Health, OpenAI for Healthcare is built to support healthcare workflows and retrieve evidence-based data. Anthropic announced a similar product, Claude for Healthcare , on Jan. 11. On the surface, ChatGPT Health is a consumer education product that is not intended to diagnose medical ailments. However, standard ChatGPT doesn't even advertise itself as a health product, yet many users will freely share sensitive personal health information with it to diagnose themselves. Some individuals will also, at times, treat LLMs as real, super-intelligent beings that can provide critical information, or participate in a human interaction like doctor-patient, rather than as datasets that predict the appropriate output (and are prone to factually incorrect outputs, or hallucinations). Safety issues associated with AI-human interactions may also have allegedly contributed to mental health crises and even deaths . All of this to say, the default high trust level many users have in LLM models could facilitate a need for controls, regulations, and regular reminders about ChatGPT's strict role as an educational tool beyond that which OpenAI has already announced. Caution Advised Corynne McSherry, legal director for digital rights nonprofit the Electronic Frontier Foundation, tells Dark Reading that even with privacy protections and promises, data breaches will happen and companies will generally comply with legal processes such as subpoenas and warrants as they come up. "If you give your data to any third party, you are inevitably giving up some control over it and people should be extremely cautious about doing that when it's their personal health information," she says. Access Now's Giandana Gigena says AI health assistants are relatively new to the market and their effect on user health has not been tested, including questions tied to the dependency these models could create or the potential harm caused by hallucinations (a problem that remains unsolved). From a policy perspective, Giandana Gigena observes that ChatGPT Health will not yet launch in the European Economic Area, Switzerland, and the UK, where stricter data regulations (such as GDPR) are in place. He calls this "concerning regarding the level of minimization, purpose limitation and overall protection the system actually offers to those who can access it." "Even if we have witnessed how companies offer different solutions and protection levels depending on which jurisdictions a user is located, the service offered in this type of assistant raises the alarm on how people are treated and the level of human rights commitment the business sector has at the end of the day," he says. Read more about: CISO Corner About the Author Alexander Culafi Senior News Writer, Dark Reading Alex is an award-winning writer, journalist, and podcast host based in Boston. After cutting his teeth writing for independent gaming publications as a teenager, he graduated from Emerson College in 2016 with a Bachelor of Science in journalism. He has previously been published on VentureFizz, Search Security, Nintendo World Report, and elsewhere. In his spare time, Alex hosts the weekly Nintendo podcast Talk Nintendo Podcast and works on personal writing projects, including two previously self-published science fiction novels. See more from Alexander Culafi