Galux Candidate Drug Selected for South Korea National R&D Program, Marking Another Step for AI Drug Discovery but Still Awaiting Validation

The longest part of drug discovery often does not lie in proposing a molecule that looks impressive, but in proving that it can act stably, safely, and reproducibly in real biological systems. The significance of an AI-designed drug from the South Korean company Galux being selected for a national new drug development program lies precisely in this: it moves AI drug discovery from a narrative about algorithmic capability toward an arena closer to public R&D resources and formal development pathways.

According to a report by the Seoul Economic Daily, Galux’s AI-designed candidate drug has been selected for South Korea’s “National New Drug Development Program.” Programs of this kind are typically intended to support R&D projects with industrial and clinical potential, helping candidate drugs cross the high barriers of funding, validation, and translation in early development. However, the publicly available summary has not disclosed the candidate drug’s specific indication, mechanism target, preclinical data, or planned development timeline.

This also means the focus of the news cannot be simplified as “AI has already designed a new drug.” In biomedicine, a candidate molecule must at minimum pass layers of evaluation including target rationale, in vitro activity, selectivity, pharmacokinetics, toxicology, and manufacturability before it may enter human trials. AI can help narrow the search space, propose structural hypotheses, or optimize molecular properties, but it cannot replace wet-lab experiments and clinical evidence.

Galux’s case reflects one direction in South Korea’s biotech policy: incorporating AI drug design into a national-level new drug development framework, rather than leaving it only in company presentations or one-off technology demonstrations. If more experimental data can be published later, outside observers will be better able to judge whether this candidate drug has improved hit rates, shortened optimization cycles, or demonstrated differentiated capabilities in a specific protein structure or disease target.

For general readers, the aspect of this type of news most easily misread is equating “selection for a program” with “a therapy may soon become available.” In fact, selection means the R&D project has passed a certain level of assessment and received support, but it is not proof of efficacy or safety. Especially in drug development, many early-stage candidates that appear promising reveal their limitations only after animal testing, toxicology studies, or Phase 1 clinical trials.

From an industry perspective, AI drug discovery is entering a second stage of competition. The first stage tested whether models could generate novel molecules; the second stage is more demanding, with the focus shifting to data quality, experimental closed loops, clinical interpretability, and regulatory acceptability. Without a clear validation pathway, even the most sophisticated model outputs are difficult to translate into drugs that can be approved, manufactured, and paid for.

Therefore, Galux’s selection for a national program is a positive signal, but it remains only an early milestone in the long course of drug development. The more substantive questions ahead will be: what disease and target this AI-designed drug addresses, whether there is preclinical evidence available for public review, and how regulators will evaluate candidate drugs designed with AI involvement. Only as those answers gradually emerge will it truly become clear how much substantive change AI can bring to new drug R&D.