\n\n
The AI‑driven drug discovery landscape is heating up on three fronts: new specialised models, virtual trial simulations that are increasingly accurate, and the first generative‑AI‑designed drugs entering late‑stage human trials. OpenAI has quietly added a dedicated life‑sciences model, gpt‑rosalind‑discovery, to its API catalog, signalling a move toward premium tools for pharma. At the same time, AI firms such as BioinvestGPT are proving the value of virtual patient simulations—accurately forecasting outcomes of high‑profile trials like Novartis’s del‑desiran failure and Moderna’s melanoma vaccine success, with five correct predictions out of six. Investment in AI drug discovery has more than doubled to $8.4 billion in 2025, though most spending still targets molecule design rather than trial prediction. The most tangible clinical milestone is Insilico Medicine’s Phase III study of rentosertib, the first generative‑AI‑discovered drug to reach this stage. While these advances promise faster, cheaper R&D, experts caution that robust clinical proof remains limited, and regulators are still formalising guidance. Watch for wider adoption of specialised models like gpt‑rosalind‑discovery, expanding virtual‑trial analytics, and the outcomes of rentosertib’s Phase III trial as the next barometer of AI’s impact on drug development. [[1]]