MedGemma: Our most capable open models for health AI development

However, generative AI models, despite their transformative potential, entail serious privacy and security risks due to the vast amounts of data involved and the opacity of their development. Moreover, there is widespread concern about models hallucinating—inventing false or misleading information when faced with insufficient data. These roadblocks are preventing the smooth implementation of generative AI…