AI-designed viruses are medicine's next hope, not its next threat

Key points

• Rutherford Appleton Laboratory, home to the world's most intense source of pulsed muons, faces closure under UK Research and Innovation budget cuts.
• Muons, subatomic particles, help scientists study magnetism, superconductivity, battery materials and much else with direct real-world applications.
• Bacteriophages, viruses that kill bacteria, have been used in medicine for more than 100 years and can already fight drug-resistant infections.
• AI is now being used to engineer bacteriophages into more effective treatments, a use that some scientists say deserves enthusiasm, not alarm.

Two letters published by The Guardian this week cut through a pair of ongoing UK debates, one about science funding and one about AI in medicine, and make a case for proportion.

What is being lost in the UK funding cuts?

Most public attention has gone to Jodrell Bank, the famous radio telescope in Cheshire. Oxford physicist Prof Stephen Blundell points to a second casualty that has received far less coverage.

The Rutherford Appleton Laboratory in Oxfordshire hosts the world's most intense source of pulsed muons. Muons are subatomic particles, heavier cousins of electrons, that scientists fire into materials to study their inner structure. The technique reveals how atoms behave in magnets, in superconductors (materials that carry electricity with zero resistance, which could transform energy grids), and in the battery chemistries the clean-energy transition depends on.

Blundell argues the facility is genuinely unique. No other country runs an equivalent machine at the same intensity. UK researchers built the global expertise around it. Losing the lab wouldn't just cost jobs; it would hand that scientific lead to other nations and leave questions about next-generation energy storage without one of their best tools.

Should people fear AI-designed viruses?

No, at least not these ones. The concern is understandable, but the framing is wrong.

Peter Forbes, writing from Muswell Hill, London, responds to recent headlines warning of "safety fears" around AI-designed viruses. The viruses in question are bacteriophages, often called phages. A phage is a virus whose only target is bacteria; it cannot infect human cells. Doctors have used phages as medicine since the early 1900s, long before antibiotics existed.

The reason they matter again now is drug resistance. Bacteria are evolving faster than new antibiotics can be developed, and phages can kill strains that no drug will touch. Natural phages are finicky, though: a phage that attacks one bacterial strain may ignore another. Engineering them to work more reliably has been a scientific pursuit for decades.

That's where AI comes in. Machine-learning models can scan enormous libraries of phage genomes, spot patterns humans would miss, and suggest modifications that make a phage more effective. When we covered the original research on 6 August, in "Scientists build the first AI-designed viruses, and experts are asking hard questions", the safety debate was already moving faster than the published evidence. Forbes's letters cuts to the heart of that: reflex alarm is the wrong response to a clinical tool with a century of use behind it.

His frustration is with framing any AI-in-biology story as a threat before the evidence warrants it.

What readers should watch for

When you see a headline about AI creating new viruses, check two things before sharing it. What does the organism actually target? Bacteriophages target bacteria, not people. And what's the existing safety record? A technology used clinically for over a century isn't a blank-slate risk.

For the funding story, watch which facilities appear in budget announcements only as line items, no public name attached. Jodrell Bank survived partly because the public knew what it was. The muon source at Rutherford Appleton may not get that chance.