It’s available for free to registered clinicians in the US, and we’re probably not far behind.
ChatGPT for clinicians has been released in the US – for free, for now – and will likely come here, adding to a number of AI-powered decision support tools being spruiked to clinicians.
Relying only on “trusted medical sources”, it was a step up from just asking Google or even regular ChatGPT, according to its maker, OpenAI, which billed it as “your secure clinical reasoning partner”.
However, OpenAI did not say how it chose its sources, whether it had any data partnerships, how long it would be without a price tag, or when it would be available in Australia, although it flagged plans “to expand access to additional countries … [in] the coming months”.
RACGP specific interest digital health and innovation deputy chair Dr Janice Tan said the college was “watching closely”.
“My position is cautiously optimistic,” she told The Medical Republic.
“The risks are real — privacy, data security, and the fact that these tools can be confidently wrong through hallucinations and embedded biases.
“Those aren’t concerns you can hand-wave away in general practice. But I’d be lying if I said I wasn’t genuinely interested in where this goes.
“What gets me isn’t just the time-saving or the productivity angle, though that matters.
“It’s the possibility that GPs might have room to think again, to be present in a consultation rather than half-distracted by paperwork.
“Burnout in general practice is at a serious level right now, and if AI can take some of that load off, it deserves serious attention.”
This article originally ran on TMR’s sister site, Health Services Daily. TMR readers can sign up for a discounted subscription.
However, regulation needed to catch up, she said.
“We need clear oversight and practical guidance so GPs aren’t navigating this on their own. Just as importantly, we need to bring the whole workforce along, not just the early adopters.
“Upskilling general practice as a profession is what gives GPs the confidence to use these tools well, and to know when not to. That’s the work ahead.”
Healthdirect CEO Bettina McMahon, who recently co-authored “Shadow AI in consumer health: the case for safe adoption of national health AI assistants”, said that as well as increasing growth, offering a free product could also be a way for Open AI to allow individual clinicians in large organisations to circumvent their employer’s AI arrangements.
“I think what they’re doing is feeding this shadow IT. Water finds a channel. People will just start using a tool that they want. We’ve seen this happen in IT for 20 years,” Ms McMahon told TMR.
OpenAI promises that its product “[s]upports clinical reasoning and document drafting, while clinicians stay in control of care decisions”.
It said it would “support evidence-based decision making” with cited answers to questions “grounded in trusted medical sources”, “advanced AI models for clinical questions”, and “secure and private” sessions where interactions wouldn’t be used to train its models.
It did not say what sources it would use specifically, and did not mention any partnerships like the ones a competitor, OpenEvidence, had with specific journals.
But GP Dr Roy Mariathas, a practising clinician steeped in digital health and AI, said the company was at the forefront of AI research.
He does not work for them, but he was the only Australian clinician involved in OpenAI’s HealthBench, a project which involved 262 physicians from 60 countries evaluating the responses of large language models in 5000 health-related conversations.
Building on that project, ChatGPT for Clinicians was released in conjunction with an “open benchmark” across three use cases: patient consultation, writing and documentation, and medical research.
Dr Mariathas told TMR the question was what architecture being used to look at those sources. “Are you able to point to those articles when they need pointing to 100% of the time?” he explained.
Related
TMR tried to find out if ChatGPT for Clinicians was launching in Australia too, but OpenAI does not provide an avenue for media queries, so we had to ask ChatGPT.
“Short answer: not yet in Australia (as of now), but likely later … expansion to Australia is expected but not scheduled publicly,” the free version of the AI model replied.
“Earliest realistic window (based on ‘coming months’ pilots): late 2026 at the absolute earliest for pilots, broader access likely later…”
ChatGPT for Health, which ChatGPT described as “consumer + light clinical support”, launched here in January.
It continued: “Australia is actively being prepared (via Health + partnerships), but: → clinician-grade rollout will depend heavily on regulation and pilots first.”
The clinician-specific AI tool already has competitors here. In addition to OpenEvidence, there is Heidi Health, Med Luma, Up to Date and Clinical Key.
Dr Mariathas suggested ChatGPT’s survival would come down to partnerships with organisations which held local data, the ability to integrate into existing clinical workflows and regulatory environment and, from a research perspective, accountability and transparency.
