AI predicts new cancer treatment and scientists confirm it

An artificial intelligence (AI) model generated a hypothesis about the behavior of cancer cells, which was later confirmed in an experiment with living cells, Google announced on Wednesday, calling the research a milestone for science.

The model, called Cell2Sentence-Scale 27b (C2S-Scale), is part of a collaboration with Yale University and is based on the Gemma family of open models.

Google explained in a statement that the model has 27 billion parameters and represents “a new frontier in the analysis of individual cells.”

"C2S-Scale generated a new hypothesis about cancer cell behavior, and we've since confirmed its prediction with experimental validation in living cells. This discovery reveals a promising new avenue for developing cancer therapies," he explained.

Google CEO Sundar Pichai highlighted the news on the social network X and emphasized that this potential is subject to "further clinical and preclinical testing."

An exciting milestone for AI in science: Our C2S-Scale 27B foundation model, built with @Yale and based on Gemma, generated a novel hypothesis about cancer cellular behavior, which scientists experimentally validated in living cells.

With more preclinical and clinical tests,…

— Sundar Pichai (@sundarpichai) October 15, 2025

Therefore, the company decided to share the model and other resources with researchers at HuggingFace and GitHub.

The goal was to find a drug that would prevent some tumors from becoming invisible to the immune system and force them to display signals that trigger an immune response through the process of "antigen presentation," which poses a challenge for cancer immunotherapy.

According to the note, signed by scientists from Google Deepmind and Google Research, the C2S-Scale model was able to generate a new and testable hypothesis, instead of “repeating known facts” , about a CK2 kinase inhibitor called silmitasertib (CX-4945).

"Although CK2 is involved in many cellular functions, including as a modulator of the immune system, no studies have reported that CK2 inhibition with silmitasertib explicitly increases antigen presentation," Google highlighted in the statement.

Google scientists tested in the laboratory a combination of silmitasertib and a low dose of interferon, an immune system signaling protein, "which resulted in a 50% increase in antigen presentation, which would make the tumor more visible to the immune system."

"The model's in silico (computational) prediction was confirmed several times in vitro (in the laboratory)," the scientists highlighted in the statement, stressing that this "first initial step" supported by AI could lead to the development of new combination drug therapies that achieve a "more robust effect."

Google noted that Yale researchers are now exploring the "discovered mechanism and testing additional AI-generated predictions in other immunological contexts."