AI Healthcare Tools Race Heats Up Among Tech Giants
259This month, OpenAI, Google, and Anthropic announced new AI tools aimed at healthcare, all within days of each other. The timing suggests a competitive push rather than coincidence. But it’s important to note that none of these tools are approved for clinical diagnosis or considered medical devices yet. They are marketed as ways to support healthcare workflows and improve processes.
New AI Offerings from OpenAI, Google, and Anthropic
OpenAI launched ChatGPT Health on January 7. It allows US users to connect their medical records through partnerships with companies like b.well, Apple Health, Function, and MyFitnessPal. Google responded with MedGemma 1.5 on January 13, an open medical AI model capable of interpreting 3D CT and MRI scans, as well as histopathology images. Anthropic followed suit on January 11 with Claude for Healthcare, providing HIPAA-compliant connectors to databases such as CMS coverage, ICD-10 coding, and the National Provider Identifier Registry.
All three companies are focusing on common healthcare pain points like prior authorization, claims processing, and clinical documentation. Although their approaches differ, they share similar technical foundations. Each uses multimodal large language models trained on medical literature and clinical data, emphasizing privacy and regulatory disclaimers. They position their tools as aids that support, not replace, clinical judgment.
Differences in Deployment and Access
The main differences lie in how these tools are accessed and deployed. OpenAI’s ChatGPT Health is designed for consumers, with a waitlist for its ChatGPT Free, Plus, and Pro users outside Europe, Switzerland, and the UK. Google’s MedGemma 1.5 is an open model available through its Health AI Developer Foundations program, which can be downloaded or run via Google Cloud’s Vertex AI. Meanwhile, Anthropic’s Claude for Healthcare integrates into existing enterprise workflows, targeting hospitals and healthcare organizations rather than individual consumers.
All three companies clarify that their AI tools are not for diagnosis or direct patient treatment. OpenAI explicitly states that ChatGPT Health is not intended for diagnosing or treating patients. Google describes MedGemma as a starting point for developers to adapt for their specific medical applications. Anthropic emphasizes that outputs should not be used to make clinical decisions, underscoring their role as supportive tools rather than direct clinical solutions.
Performance Benchmarks and Clinical Readiness
While these new AI tools show impressive benchmark results, there is still a gap between test performance and clinical use. Google reports that MedGemma 1.5 achieved 92.3% accuracy on MedAgentBench, a Stanford medical task benchmark, up from 69.6% with the previous version. These improvements are promising but don’t yet mean the tools are ready for real-world medical deployment. The same is true for OpenAI and Anthropic’s offerings, which are still in the development and testing phases.
Overall, the race among these tech giants highlights how AI is increasingly being positioned to support healthcare workflows. But for now, these tools serve as developer platforms or supportive systems rather than direct diagnostic products. The focus remains on improving administrative tasks and clinical documentation while ensuring privacy and regulatory compliance.
What do you think?
It is nice to know your opinion. Leave a comment.