OpenAI and Anthropic launch new health-focused AI tools

Last week, OpenAI announced ChatGPT Health, which allows users to discuss health in a private setting where messages are not used as training data. This week, Anthropic announced its own health product, Claude for Healthcare. The launches represent a push by artificial intelligence developers into the medical sector, aiming to address administrative burdens while facing immediate scrutiny over data privacy and clinical accuracy.

Industry experts are divided on the utility of these tools. While over 230 million people already talk to ChatGPT about their health each week, clinicians warn that AI chatbots have a persistent problem with hallucinations. According to Vectara’s Factual Consistency Evaluation Model, OpenAI’s GPT-5 is more prone to hallucinations than many Google and Anthropic models. Dr. Sina Bari, a practicing surgeon and AI healthcare leader at data company iMerit, noted that ChatGPT can lead patients astray with faulty medical advice. “I recently had a patient come in, and when I recommended a medication, they had a dialogue printed out from ChatGPT that said this medication has a 45% chance of pulmonary embolism,” Dr. Bari said.

Other experts raise concerns about data transfer. Itai Schwartz, co-founder of data loss prevention firm MIND, questioned how regulators would approach medical data transferring from HIPAA-compliant organizations to non-HIPAA-compliant vendors. HIPAA is the US health data privacy law. However, Andrew Brackin, a partner at Gradient, argued that because health queries are already one of ChatGPT’s largest use cases, building a private, secure, and optimized version makes sense.

Some experts argue that AI is better suited for provider-side tasks rather than patient-facing chatbots, particularly given the current shortage of patient access in the American market. Dr. Nigam Shah, a professor of medicine at Stanford, chief data scientist for Stanford Health Care, and co-founder of Atropos Health, pointed out that wait times for a primary care doctor at any health system are currently three to six months.

To address this, developers are targeting administrative tasks, which medical journals report often consume about half of a primary care physician’s time. Stanford is developing ChatEHR, a software integrated into electronic health records (EHR) systems to help clinicians retrieve patient information more efficiently. Dr. Sneha Jain, an early tester of ChatEHR, noted that making electronic medical records more user-friendly allows physicians to spend less time searching for information and more time talking to patients. Meanwhile, Anthropic’s Claude for Healthcare could reduce time spent on administrative tasks like prior authorization—the insurance approval process for medical procedures or medications. Anthropic Chief Product Officer Mike Krieger stated that cutting 20, 30 minutes out of each prior authorization case represents a dramatic time savings for clinics.

Why it matters

As AI companies launch healthcare-specific products, industry experts are debating the tension between the potential for AI to reduce administrative burdens for doctors and the risks of patient data privacy and AI hallucinations.