theSIGNAL SCIENCE
21 October 2026
"The real problem is not whether machines think but whether men do."
Science

Could AI Change the Stakes on Bioweapons?

Britain's spy chief says artificial intelligence could raise the danger of biological weapons, and a leading health-security expert explains why governments are uneasy.

Once a year, the head of MI5 steps up to a podium and tells Britain what keeps him awake. This year, Sir Ken McCallum chose a phrase that sounds almost like a gambler's warning: artificial intelligence, he said, has the potential to "change the stakes" on bioweapons. Which stakes? And how?

To find out, the Guardian's Ian Sample turned to Dr Tom Inglesby, director of the Johns Hopkins Center for Health Security, based within the Bloomberg School of Public Health in the United States. His field is an odd one. Inglesby and his colleagues spend their days imagining outbreaks that nobody wants to imagine, so that governments can prepare for them. The worry, broadly, is that AI tools might lower the barriers to dangerous biology.

They could make specialist knowledge easier to reach, or speed up work that once demanded years of training. The concern covers existing threats, and possibly new ones too. Notice the hedging, though. Officials speak of "potential", not of certainty.

Photo: Julia Koblitz / Unsplash
A researcher works in a high-security laboratory where scientists study dangerous pathogens and plan defences against them.
Discuss
  • Why might a spy chief, rather than a scientist, be the one to warn about AI and biology?
  • The article stresses hedged words like 'potential'. Why does careful language matter when discussing possible catastrophic risks?
  • How could the same AI tools both increase biological danger and help detect outbreaks earlier?
Science

Your Face May Predict Cancer Survival

Can a selfie reveal how long you might live? Boston researchers built an AI, trained on over 40 million faces, that estimates biological age and cancer survival from one photograph. Radiotherapy patients who looked five-plus years older than their real age faced a 21% higher death risk. Those who looked younger had a 17% lower one. Strangely, each tool reads different facial regions.…
  • Would you share a selfie with your doctor if AI judged your survival?
∞
Science

The Asteroid That Cooled Too Fast

In 2018, NASA's OSIRIS-REx arrived at Bennu expecting sandy beaches. It found boulders, boulders everywhere. Yet Bennu cooled as quickly as sand, which solid rock should never do. So what was going on? Samples returned to Earth, and studied by Andrew Ryan's team in Arizona, gave the answer. The rocks are highly porous, and they are also laced with tiny cracks.…
  • If cracks can fool telescopes about an asteroid, what else might we be misreading?
GLOBAL · Science
Who's paying whom? The circular money machine behind the AI boom (or bubble)
ASIA · Science
Japan to launch historic round-trip mission to Mars moon
WORLD · Science
Mission set to blast off for Mars to grab a piece of its moon
AMERICAS · Science
Having a baby is too expensive, 4 in 5 parents say
theSIGNAL IN THE LAB
1VOCABULARY
change the stakeslower the barriers
direction of travelsafeguardsscreen orders
porouslaced with
2GRAMMAR FOCUS
Modal verbs of deduction and speculation (must have, cannot have, might have)
Use 'must have + past participle' when you are almost certain something happened, 'cannot/can't have' when you are almost certain it did not, and 'might/may/could have' when it is possible but uncertain. Use 'must be' or 'cannot be' for present deductions.
must have · cannot have · might have · could have · must be · cannot be · may have
  1. The head of MI5 chose a dramatic phrase, so he been worried about how seriously people would take the threat.
  2. Officials speak only of 'potential', which suggests that a chatbot yet helped anyone to brew a plague.
  3. Dr Inglesby and his colleagues spend their days imagining outbreaks, so their work a strange and uncomfortable one.
  4. Because AI tools can speed up work that once demanded years of training, a determined individual gained dangerous knowledge much faster than before.
  5. The Boston researchers trained their AI on over 40 million faces, so the project required enormous computing power.
  6. A patient who looks five years older than their real age been through a hard period of illness, but the AI cannot say for certain what caused it.
3PHRASAL VERBS
Match the phrasal verb to its definition. All six appear in today’s articles.
  1. step up (to a podium)
  2. turn to (someone)
  3. spend (their days) imagining
  4. speed up (work)
  5. find out
4CRITICAL THINKING
Two of today's stories describe technologies that reveal hidden information: an AI that reads age and survival odds from a face, and a study of Bennu that exposed cracks no one could see from orbit. Who should be allowed to access this kind of hidden information about a person, and what could go wrong if employers or insurers did? Consider whether a prediction that is statistically useful for groups is fair to use on individuals.
5CREATIVE · HEADLINES
Write a headline for the top story in each of the following styles. One line each, no explanation:
  • TABLOID NEWSPAPER
  • LUXURY MAGAZINE
  • ACTIVIST BLOG
6WRITING
Imagine you work for a government agency that has just been given a budget to prepare for risks from new technologies. Write a short memo (120-150 words) arguing which single risk from today's stories deserves the most urgent attention, using at least two specific details to support your choice.
7DEGREES OF EXTREMITY
Complete each ladder from mild to strong.
  • worrying→→
  • unusual→→
  • surprised→→
  • important→→
  • quick→→
  • tiny→→
8SPEAKING
  1. Which would you trust more to explain risk: a scientist, a politician, or an AI?
  2. Should an AI estimate your biological age from a photo, and who should see it?
  3. What everyday technology do you rely on that you barely understand?
  4. Describe a time when something looked simple but turned out to be complicated.