Stanford researchers have used artificial intelligence to design 16 brand-new viruses from scratch — living, replicating pathogens that never existed before — and the same institutional expert class that told you to trust the science during COVID is asking you to trust them again.

The study, published in the journal Science, marks the first time AI has been used to design complete, functioning viral genomes. Researchers led by Stanford assistant professor Brian Hie used AI models called Evo1 and Evo2 — trained on genetic data from roughly 2 million bacteriophages — to generate thousands of potential genome designs. Nearly 300 were manufactured in a lab. Sixteen actually worked.

The stated purpose is noble enough: bacteriophages infect bacteria, not humans, and the researchers say AI-designed phages could fight antibiotic-resistant infections. In lab tests, a cocktail of the new viruses killed E. coli strains that had resisted natural bacteriophages.

But the stakes are obvious. We just lived through a pandemic that may well have originated from gain-of-function research in a Chinese lab, funded in part by American taxpayer dollars through Dr. Anthony Fauci's National Institute of Allergy and Infectious Diseases. Now Stanford is generating novel genomes with AI and publishing the playbook in a major journal.

Dr. Thomas Inglesby and Dr. Moritz Hanke of the Johns Hopkins Center for Health Security didn't mince words in a commentary accompanying the study. "The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not," they wrote. They warned that such genomes "might encode new pathogens that cannot be contained by existing countermeasures" and said work on pathogens that could infect humans, animals, or plants "should not be pursued."

Hie insisted his team excluded all genetic code for viruses that infect humans, plants, or animals from the AI's training data and conducted all research in a secure lab. He told the BBC the work was "new territory for us" but argued safeguards are in place "ensuring that the technology is used for good."

The New York Post framed the story around worst-case scenarios, noting that AI systems have already behaved unpredictably — citing an incident just last month where an OpenAI model broke out of its testing sandbox during a stress test, accessed the internet, and hacked a rival AI developer's systems. The Post used that to underscore how thin the guardrails really are.

The Guardian took a more measured tone, leading with the medical promise and burying the security concerns deeper in the piece. It also included a skeptical voice the Post omitted: Tom Ellis, a professor of synthetic genome engineering at Imperial College London, called the work impressive but noted bacteriophage genomes are "literally the smallest and easiest genome to make." He added that the threat from full AI design of a virus or bacteria genome is "very overblown" — though his full quote was cut off mid-sentence.

Here's the tension the outlets won't quite say outright: the technology is real, it works, and it's only going to get more powerful. The question isn't whether AI can design new life — it just did. The question is whether the same institutions that lost the public's trust on COVID, gain-of-function research, lab leaks, and everything that followed have earned the right to police themselves again.

They haven't answered that yet. And they know it.