China Approves First AI-Designed Drug in 3.5 Years

Finding new medicines just got dramatically faster, and China's breakthrough shows how artificial intelligence is turning decades of drug development into years.

China's National Medical Products Administration approved Mprosevir, an antiviral treatment for mild to moderate COVID-19 in adults. The drug earned China's Class 1 designation, reserved for completely new medicines never before marketed anywhere in the world.

What makes this approval historic isn't just the drug itself. It's the first time regulators anywhere have approved a small-molecule drug designed using DNA-encoded library technology paired with artificial intelligence.

The development team from Westlake University, Westlake Laboratory, and Westlake Pharmaceuticals started with a library of 49 billion chemical compounds. Traditional methods would require armies of researchers and years to screen even a fraction of that number.

Instead, the DNA-encoded library identified over 100 promising candidates in the first stage. Then artificial intelligence took over, analyzing the candidates and narrowing the list to just nine molecules in days, not years.

Six of those nine showed higher activity in testing, proving the AI made accurate predictions. After evaluating toxicity and other key properties, researchers selected molecule WLU6937 as their lead candidate.

The team spent two more years optimizing the molecule, improving its effectiveness while reducing toxicity. AI continued guiding decisions through cell experiments, mouse studies, and clinical trials.

The Ripple Effect

This breakthrough arrives 34 years after scientists first conceived of DNA-encoded library technology. The wait for real-world application is finally over, and the impact reaches far beyond one drug.

Pharmaceutical companies typically spend 10 to 15 years developing a single new medicine. The costs run into billions of dollars, with most candidates failing before reaching patients.

Mprosevir's journey from initial discovery to completed clinical trials took three and a half years. That timeline represents a fundamental shift in what's possible when human expertise combines with machine learning.

The development team calls their work proof that artificial intelligence is already an effective and feasible tool in drug development. That assessment feels modest given the results.

Researchers worldwide now have a roadmap for integrating AI into their own drug discovery programs. The technology can screen billions of compounds, predict which candidates will succeed, and guide optimization, all while reducing costs and saving time.

Every year saved in development means patients get access to new treatments sooner. Every dollar saved can fund research into diseases that affect smaller populations, conditions that pharmaceutical companies often ignore because traditional development costs too much.

China's approval proves AI-assisted drug design works in practice, not just theory, opening the door for medicine to move at the speed of innovation.