OpenAI has launched a new ChatGPT feature in the US that can analyse people's medical records to provide better answers, but campaigners warn it raises privacy concerns.

The firm wants people to share their medical records along with data from apps like MyFitnessPal, which will be analysed to give personalised advice.

OpenAI said conversations in ChatGPT Health would be stored separately from other chats and would not be used to train its AI tools, as well as clarifying it was not intended to be used for "diagnosis or treatment".

Andrew Crawford, of US non-profit the Center for Democracy and Technology, said it was "crucial" to maintain "airtight" safeguards around users' health information.

It is unclear if or when the feature may be introduced in the UK.

"New AI health tools offer the promise of empowering patients and promoting better health outcomes, but health data is some of the most sensitive information people can share and it must be protected," Crawford said.

He said AI firms were "leaning hard" into finding ways to bring more personalisation to their services to boost value.

"Especially as OpenAI moves to explore advertising as a business model, it's crucial that the separation between this sort of health data and memories that ChatGPT captures from other conversations is airtight," he said.

According to OpenAI, more than 230 million people ask its chatbot questions about their health and well-being every week.

In a blog post, it said ChatGPT Health had "enhanced privacy to protect sensitive data".

Users can share data from apps like Apple Health, Peloton and MyFitnessPal, as well as provide medical records, which can be used to give more relevant responses to their health queries.

OpenAI said its health feature was designed to "support, not replace, medical care".

'Watershed moment'

Generative AI chatbots and tools can be prone to generating false or misleading information, often stating this in a very matter-of-fact, convincing way.

But Max Sinclair, chief executive and founder of AI marketing platform Azoma, said OpenAI was positioning its chatbot as a "trusted medical adviser".

He described the launch of ChatGPT Health as a "watershed moment" and one that could "reshape both patient care and retail" - influencing not just how people access medical information but also what they may buy to treat their problems.

Sinclair said the tech could amount to a "game-changer" for OpenAI amid increased competition from rival AI chatbots, particularly Google's Gemini.

The company said it would initially make Health available to a "small group of early users" and has opened a waitlist for those seeking access.

As well as being unavailable in the UK, it has also not been launched in Switzerland and the European Economic Area, where tech firms must meet strict rules about processing and protecting user data.

But in the US, Crawford said the launch meant some firms not bound by privacy protections "will be collecting, sharing, and using peoples' health data".

"Since it's up to each company to set the rules for how health data is collected, used, shared, and stored, inadequate data protections and policies can put sensitive health information in real danger," he said.