Mind & Body · Saturday, 31 October 2026

90% of 5,500 US pupils aged 9 to 17 have used an AI chatbot, and nearly two-thirds use one both for homework and when stressed or lonely, a survey finds

23 stories 36 sources

Signals on this beat

The bar and the word are our reading of how hard each one is pushing, and the arrow is where it is heading.

Limits on children's apps Building↑

OpenAI said it will limit weapons information for teens and tell parents when a teen's account is paused. [5]
Meta settled with a second teenager who said Instagram and Facebook were built to be addictive. [7]

What people say versus what is measured Steady→

Students prone to worry reported less sleep than their wrist monitors recorded. [24]
About 28% of US adults in a survey admitted exaggerating or inventing symptoms to get antibiotics. [20]

Weight and eating High↑

New cases of anorexia and bulimia worldwide rose 71% between 1990 and 2023. [10]
MPs said England's plan to end its obesity epidemic is doomed to fail without a rethink. [11]

Stress reaching the heart Building↑

Heart patients had 77% more irregular heartbeats around the 2016 US election. [14]
Heart attacks rise in the days after the spring clock change, and less after the autumn one. [15]

The stories

23 stories on this beat today.

Every story told the same way: what happened, who is involved, why it matters, how it works, the truth and the good news.

01
90% of 5,500 US pupils aged 9 to 17 have used an AI chatbot, and nearly two-thirds use one both for homework and when stressed or lonely, a survey finds

90%

of 5,500 US pupils aged 9 to 17 have used an AI chatbot

nearly two-thirds use one both for schoolwork and when stressed or lonely [1]

12%

of US teens told Pew in 2025 they used chatbots for emotional support

Pew asked 1,458 teens aged 13 to 17, a narrower age range than JED [3][1]

13

the minimum age on most popular chatbots

nearly three in four children have used one by that age [1]
What happened
- A survey of 5,500 US pupils aged 9 to 17 found 90% had used AI chatbots. [1]
- Nearly two-thirds use chatbots both for schoolwork and when stressed, lonely or overwhelmed. [1]
- A second survey, by Common Sense Media, found nearly half of 9-to-12-year-olds have used one. [1]
- By age 13, nearly three in four children had used chatbots, it found. [1]
- Most popular chatbots set 13 as their minimum age. [1]
- In the JED survey, 58% said they use chatbots for help with schoolwork. [2]
- A similar number turn to chatbots when they feel sad, stressed, lonely or overwhelmed. [2]
- JED also asked pupils about their mood, loneliness and any past thoughts of suicide. [1]
- Pupils who used chatbots when distressed reported worse mental health. [1]
- They also reported more past suicidal thoughts and suicide attempts. [1]
- JED's research head, psychologist Kayla Anderson, says the data cannot show which causes which. [1]
- Her team followed middle schoolers in one US state who use chatbots. [1]
- Pupils who used chatbots for company grew lonelier over time, her study, not yet published, found. [1]
- Her team found the same pattern in a study of young adults. [1]
- A team writing in the journal Nature Mental Health built a checklist from clinicians' advice. [4]
- It scores how chatbots answer teenagers who describe anxiety or depression. [4]
- They tested 4 chatbots, ChatGPT-5, ChatGPT-4, Claude 4.5 and Gemini 2.5, with made-up teenage users. [4]
- Answers covered more of the checklist for anxiety than for depression. [4]
- They also covered more for moderate symptoms than for severe ones. [4]
- ChatGPT will give teens less practical information about weapons, even when a question seems harmless. [5]
- For all users, ChatGPT will be able to end chats that seem fixated on violence. [5]
- OpenAI says it now guesses users' ages better and moves under-18s into a teen version. [5]
- It plans to tell parents if a teen's account is paused for repeated talk of violence. [5]
Who is involved

The JED Foundation, a US group that campaigns on young people's mental health, ran it. [1]
Common Sense Media is a US group that campaigns for children's safety online. [1]
Anne Maheux is a developmental psychologist at the University of North Carolina. [1]
On Friday OpenAI, the company that makes ChatGPT, said it will tighten its rules for teens. [5]
- The JED Foundation a US group that campaigns on young people's mental health; it surveyed 5,500 pupils
- Common Sense Media a US group that campaigns for children's safety online; it surveyed children from age 9
- Anne Maheux a developmental psychologist at the University of North Carolina; her study found pupils grew lonelier
- OpenAI the company that makes ChatGPT; it announced new limits for teens on Friday
How it unfolded

In autumn 2025 the Pew Research Center, a US research group, surveyed 1,458 teens. [3]
Pew asked only teens aged 13 to 17. [3]
Just 12% of them said they used chatbots for emotional support or advice. [3]
More than half used chatbots to search for information or help with schoolwork. [3]
1. Autumn 2025 Pew surveys 1,458 US teens; 12% use chatbots for emotional support [3]
2. 28 Oct NPR reports the JED survey of 5,500 pupils [2]
3. 30 Oct Common Sense Media's survey is reported, and OpenAI tightens its teen rules [1][5]
4. Coming weeks OpenAI's new limits start to roll out [5]
Why it matters

Children now ask chatbots for comfort about as often as for homework help. [1]
So software is answering children at the moments they feel stressed, lonely or anxious. [1]
How it works

Psychologists call it coping: getting through everyday stress without being knocked over by it. [1]
Maheux thinks a chatbot eases stress in the moment but stops young people practising coping. [1]
Why what people report about themselves can differ from what a device records

Psychologists call it self-report bias: answers pass through memory, mood and the wish to look good, so researchers check them against devices and records.

The twist

A survey answer is a person's memory of what happened, so it can carry their worry, their shame or their wish to look good.

The picture

6.74 hoursAsked directly
6.70 hoursWrist monitor
7.40 hoursFrom usual bedtimes
Three ways of measuring the same 134 students' sleep. The direct answer and the monitor nearly agree; working it out from remembered bedtimes gives more.
Step by step
1. Some things can only be learned by asking people
2. Each answer passes through memory, mood and the wish to look good
3. So reported sleep, symptoms or habits can drift from what happened
4. Researchers compare answers with a device or a record, and ask in private
5. Where the two differ, the gap itself becomes a finding
The same force, elsewhere today

Where this chain is also running, in today's other stories.
- Worriers sleep longer than they say
  
  Students prone to worry said they slept less, while the monitors on their wrists recorded no less sleep.
- 28% admit lying to get antibiotics
  
  The figure comes from people admitting a lie in a survey, and some who lied may not have said so.
- Elections and the heart
  
  The 77% rise was counted by implanted heart devices, so it did not rest on what patients remembered feeling.
- Children using chatbots for comfort
  
  Every link JED found between chatbot use and poor mental health rests on what pupils said about both.
Where you've seen this

Food diaries

people tend to leave out snacks when they write down what they ate

Exercise surveys

remembered workouts often run longer than a step counter shows

Polls about voting

more people say they voted than official turnout records show

The catch

A device has its own errors: a wrist monitor can count lying still and awake as sleep, and only the person can report a feeling like loneliness.

And the whole of it

A doctor hears the patient's account, a researcher reads the survey, and a device records only what it can sense. Each sees part of what happened, and the decisions in today's stories rest on putting those parts together.
The truth
- What is really going on Most US pupils now use AI chatbots, and nearly two-thirds also use them when stressed or lonely, while nobody can yet say whether that helps or harms them. [1]
- Why it works on people A big share like 90% feels like proof of harm, though it counts only how many pupils have used a chatbot. [1]
- Children under 13 use chatbots, though most of the apps set 13 as the minimum. [1]
- Nobody yet knows whether chatbot comfort helps lonely pupils or makes them lonelier. [1]
- Who pays Children under 13 — Nearly three in four have used chatbots by 13, the minimum age most of them set. [1]
- Nobody knows yet
  Whether chatbots make lonely pupils lonelier, or lonely pupils seek out chatbots.
  
  JED's data cannot tell which comes first, and the University of North Carolina study pointing both ways is not yet published. [1]
- Nobody knows yet
  How many teens really use chatbots for comfort.
  
  Pew found 12% of 13-to-17-year-olds in autumn 2025. [3] JED found nearly two-thirds of pupils aged 9 to 17 use them when stressed, lonely or overwhelmed. [1] The two asked different ages at different times.
What comes next

The new limits will start rolling out over the next few weeks, OpenAI says. [5]

02
Warnings come off three more menopause patches
What happened
- The FDA said on Friday 3 more menopause patches will lose their black-box warnings. [6]
- The warnings were about heart disease, breast cancer and dementia. [6]
- Again it did not ask its outside expert committee. [6]
- Asked for the evidence, officials said it could be provided later. [6]
- Since February, patients in many areas have struggled to find their usual estrogen patches. [6]
- Production of estrogen patches has doubled in a year, a US health official said. [6]
- The FDA says risks look different for many healthy women under 60. [6]
Who is involved

The FDA, the US Food and Drug Administration, sets the warnings printed on medicines. [6]
- The FDA US Food and Drug Administration
- Reshma Ramachandran Yale School of Medicine
How it unfolded

The FDA took the same warnings off six menopause treatments in February. [6]
1. February six treatments
2. Friday three more patches
3. Friday 3 more patches
4. Next the remaining hormone treatments
Why it matters

Women choosing hormone treatment will now see fewer warnings on three more patches. [6]

How it works

In pharmacology, the route a drug takes into the body changes its risks. [6]
Patches skip the liver on their way into the blood, so doctors see less risk. [6]
The truth
- Yale researcher Reshma Ramachandran says removing them without clear evidence underplays real harms. [6]
- Women who have had hormone-sensitive cancers face particular risks, she says. [6]
- Who gains Makers of estrogen patches — Demand rose sharply after the February label change, and making of the patches has doubled in a year. [6]
- Who pays Women who have had hormone-sensitive cancers — Three more patches lose warnings about breast cancer, and a Yale researcher says such women face particular risks. [6]
- Nobody knows yet
  What evidence the FDA used to take the warnings off the patches.
  
  It did not consult its outside expert committee, and officials said the evidence could be provided later. [6]
Good news
- Makers of estrogen patches in the US have doubled production in a year, a health official said, as women struggled to find them.
What comes next

It is now working to change the labels of the remaining hormone treatments. [6]
03
Meta settles before a jury can rule
What happened
- Meta settled on Friday with a teenager suing over Instagram and Facebook, her lawyers said. [7]
- She said the apps were built to be addictive, causing her depression and disordered eating. [7]
- Her case was picked as a test case from thousands of similar US lawsuits. [7]
- Meta had settled with the trial's other teenager days before. [7]
Who is involved

Meta is the company that owns Instagram, Facebook and WhatsApp. [8]
- Meta owns Instagram, Facebook and WhatsApp
- Snap owns Snapchat
- YouTube found negligent
How it unfolded

They blame features like autoplay and endless scrolling. [7]
In March, the first such trial, in Los Angeles, found Meta and YouTube negligent. [7]
In August, Meta agreed settlements worth up to $18 billion with nearly every US state. [7]
1. Days before settles with the first teenager
2. Friday settles with the second
Why it matters

A verdict in a test case can shape how thousands of other cases are settled. [7]

How it works

In law, a test case goes first to show how juries see similar claims. [7]
The truth
- What is really going on The first limits are coming from companies and courts: OpenAI tightened its teen rules on Friday, and Meta settled a second teenager's addiction case before a jury could rule. [5][7]
- Meta settled before a jury could rule on how its apps were designed. [7]
- The companies deny building their apps to addict young people. [7]
- Who gains Meta — Settling with both teenagers before a verdict kept a jury from ruling on Instagram's design in this test case. [7]
- Nobody knows yet
  Whether the Los Angeles trial now reaches a verdict.
  
  Meta has settled with both teenagers, and the trial is expected to go on against Snap alone. [7]
What comes next

So the trial is expected to go on against Snap, which owns Snapchat, alone. [7]
04
Eating disorders up 71% worldwide
What happened
- New cases of anorexia and bulimia worldwide rose 71% from 1990 to 2023, a study finds. [10]
- In 2023, about 18.1 million people had one of the two disorders, it estimates. [10]
- In her small study, 30% of people with eating disorders had used weight-loss drugs like Ozempic. [10]
- About 10% appeared to have misused them. [10]
Who is involved

Anorexia is defined by a fear of gaining weight; in bulimia, people binge then purge. [10]
Ruhai Bai's team at a children's hospital in Nanjing, China, published it in JAMA Psychiatry. [9]
Psychologist Cheri Levinson says the two are under half of all eating disorders. [10]
- Ruhai Bai Children's Hospital of Nanjing Medical University
- Cheri Levinson clinical psychologist, University of Louisville
Why it matters

The drugs cause fast weight loss, which could push people with anorexia further into danger. [10]

How it works

Clinical psychologists say cutting food can itself trigger an eating disorder in people already prone. [10]
The truth
- Children as young as 11 have bought such drugs through TikTok, Levinson says. [10]
- The study did not look at why cases rose; social media and lockdowns are suspects. [10]
- Good figures stop in 2023, so nobody can yet measure any rise since. [10]
- Nobody knows yet
  How far eating disorders have risen since 2023.
  
  Good figures stop in 2023, and one psychologist says the true burden may be three times the estimate. [10]
What comes next

Guidelines out this month tell doctors to check for eating disorders before prescribing these drugs. [10]
05
MPs: England's obesity plan is doomed
What happened
- England's plan to end its obesity epidemic within a decade is 'doomed to fail', MPs say. [11]
- Basic questions, like how progress will be measured, are still unanswered, the MPs said. [11]
- About 30% of adults in England live with obesity. [11]
- Close to a third of children are overweight or living with obesity. [11]
- The MPs want a cabinet committee on obesity, chaired by the prime minister. [11]
- England's health department says it already limits junk food adverts before 9pm. [11]
Who is involved

The MPs sit on the Commons health committee, chaired by Liberal Democrat Layla Moran. [11]
- Layla Moran chairs the Commons health committee
- A cabinet committee chaired by the prime minister
How it unfolded

The aim was set in England's 10-year health plan more than a year ago. [11]
1. Now the MPs' report
2. Within 3 months a full plan due
Why it matters

Without ways to measure progress, MPs cannot hold ministers to their own promise. [11]

How it works

In public health, preventing obesity and treating it are different jobs with different tools. [11]
The MPs say mixing the two up has muddled every government's plan. [11]
The truth
- Moran says a real plan means standing up to the food industry on adverts and labels. [11]
What comes next

They want ministers to bring a full plan to parliament within 3 months. [11]
06
Police in Northern Ireland step back from welfare checks
What happened
- From 1 November, Northern Ireland's police will stop most mental health welfare checks. [12]
- Police will still attend if there is a crime, or a risk of harm or death. [12]
- Only 3% of calls about people possibly in danger involve a crime, police say. [12]
- Callers still dial 999, and handlers may pass calls to health or social care staff. [12]
- Ross Irvine, of the Royal College of Psychiatrists, says 30% of mental health posts are empty. [12]
Who is involved

The force, the PSNI, answers about 30,000 mental health calls a year. [12]
How it unfolded

The policy, Right Care Right Person, has run in parts of England and Wales since 2023. [12]

Why it matters

Calls the police drop will land on health services already short of staff. [12]

How it works

Emergency services call this triage: sending each call to the service best able to help. [12]
In one US study, sending mental health workers with police cut forced psychiatric detentions 16.5%. [13]
The truth
- Social workers say the system lacks the staff to take over the calls safely. [12]
- Who gains Police in Northern Ireland — Officers stop most welfare checks from 1 November, and only 3% of those calls involved a crime. [12]
- Who pays Northern Ireland's mental health services — They take on calls the police drop while 30% of their posts are vacant. [12]
- Nobody knows yet
  Whether Northern Ireland's health services can take the calls police drop.
  
  A psychiatrists' leader says 30% of posts are empty, and social workers say the change is too quick. [12]
07
Elections and the heart
What happened
- Elections can strain the heart, US researchers say, ahead of Tuesday's midterm elections. [14]
- Around the 2016 US election, heart patients had 77% more irregular heartbeats than usual. [14]
- Their own implanted heart devices counted the beats, in North Carolina. [14]
- Democrats and Republicans had a similar risk. [14]
- After the 2020 vote, heart attack admissions in Southern California rose over the next 5 days. [14]
Who is involved

Lindsey Rosman, a cardiology researcher at Northwestern University, led that study. [14]
- Lindsey Rosman cardiology researcher, Northwestern University
How it unfolded
1. Election day the 2020 vote
2. Next 5 days more heart attack admissions
Why it matters

Nearly half of US adults have some heart or blood vessel disease, including high blood pressure. [14]

How it works

In cardiology, sudden stress is a known trigger of abnormal heart rhythms. [14]
Stress raises adrenaline, a chemical that speeds the heart, and weakens the calming vagus nerve. [14]

The 77% rise was counted by implanted heart devices, so it did not rest on what patients remembered feeling.
The idea, in story 01: Why what people report about themselves can differ from what a device records →
The truth
- Both studies show a link, not proof that elections cause heart problems. [14]
What comes next

Rosman next plans a larger study of the 2020 and 2024 elections. [14]
08
The US clocks go back on Sunday
What happened
- Most of the US sets its clocks back an hour at 2am on Sunday. [15]
- That returns the country to standard time until March. [15]
- Sleep scientists say the autumn change is easier on the body than the spring one. [15]
- Heart attacks rise in the days after March's change, but less after November's. [15]
- Some research finds more strokes after both changes. [15]
- Doctors' groups, including the American Medical Association, oppose permanent daylight saving time. [15]
Who is involved
- American Medical Association opposes permanent daylight saving time
How it unfolded

In July the US House voted 308 to 117 to make daylight saving time permanent. [18]
The bill has stalled in the US Senate, though President Trump has pressed for it. [17]
1. Sunday standard time
2. March daylight saving time
Why it matters

Standard time brings earlier morning light, which suits the body clock better, experts say. [15]

How it works

The brain's clock is about 40,000 nerve cells at the base of the brain. [16]
Left alone, it runs a day of roughly 24 hours and 12 minutes. [16]
Morning light pulls it back into step with the sun each day. [16]
The truth
- Sleep gadgets can cost thousands, while bright days and dark nights cost nothing. [16]
Good news
- In England, the week after the autumn change had fewer recorded cases of anxiety and depression. [19]
- In England, the week after the autumn clock change brought fewer recorded cases of anxiety, depression and heart disease, in a study of 683,809 people.
09
28% admit lying to get antibiotics
What happened
- About 28% of nearly 960 US adults admitted exaggerating or inventing symptoms to get antibiotics. [20]
- The survey was published on Friday in the journal JAMA Network Open. [20]
- Most lies were about cold-like symptoms such as fever, cough and sore throat. [20]
- Among those who lied, 77% said it had worked at least once. [20]
- In a second survey, 27% of 236 parents and carers lied about a child's symptoms. [20]
How it unfolded

Antibiotics do not treat infections caused by viruses. [20]
US infections from drug-resistant germs rose from 2.8 million in 2017 to 9.7 million in 2023. [20]

Why it matters

Every needless course gives bacteria more pressure to evolve resistance. [20]

How it works

Microbiologists call this selection pressure: each course pushes bacteria to evolve ways to survive. [20]

The figure comes from people admitting a lie in a survey, and some who lied may not have said so.
The idea, in story 01: Why what people report about themselves can differ from what a device records →
The truth
- Doctor Jeffrey Linder notes about 80% of those who lied still got a prescription. [20]
- He says many doctors are simply worn down by what they see as patients' demands. [20]
- The true share is unknown: some people who lied may not have said so. [20]
- Who pays People who later need an antibiotic that works — Needless prescriptions push bacteria to resist, and US resistant infections rose from 2.8 million to 9.7 million. [20]
- Nobody knows yet
  How common lying for antibiotics really is.
  
  The surveys were not fully representative of the US, and some people who lied may not have said so. [20]
10
Late clot removal helps, with more deaths
What happened
- Removing a stroke clot 24 to 72 hours after it began still helped, a trial finds. [22]
- The trial ran at 35 stroke centres in China, with 336 patients. [21]
- All had brain scans showing tissue that could still be saved. [22]
- After 90 days, treated patients had recovered more function than those on usual care. [22]
- But 6% of treated patients died within 90 days, against 1.8% on usual care. [22]
- Bleeding in the brain with symptoms hit 4.8% of treated patients and none of the others. [22]
- It was published in the New England Journal of Medicine. [21]
Who is involved

The procedure, thrombectomy, pulls the clot out through a thin tube in a blood vessel. [21]
How it unfolded

It had been proven to help only up to 24 hours after a stroke began. [21]

Why it matters

Many stroke patients reach hospital hours or even days late, when the options shrink. [21]

How it works

Neurologists say 'time is brain': about 1.9 million nerve cells die each untreated minute. [21]
Late treatment worked only where scans showed brain tissue still worth saving. [22]
The truth
- With only 13 deaths in total, chance cannot be ruled out for that gap. [21]
- Doctors must weigh better recovery against a higher risk of bleeding and death. [22]
- Nobody knows yet
  Whether late clot removal saves more people than it harms.
  
  Treated patients recovered better but died more often, and with only 13 deaths in total chance cannot be ruled out. [21][22]
11
Parents bring the second measles shot forward
What happened
- More US parents are giving children their second measles shot early, the Guardian reports. [23]
- US measles cases this year are about twice last year's. [23]
- But the shot works when given 4 weeks after the first, which the CDC allows. [23]
- Germany, the UK and France already advise giving it soon after the first. [23]
- Ages 4 to 6 were picked mostly to fit school starts and save doctors' visits. [23]
Who is involved

The CDC, the US health agency, usually advises the second dose at ages 4 to 6. [23]
- The CDC US public health agency
- Catherina Becker scientist in Germany
How it unfolded

Germany moved its second dose to 4 weeks after the first in 2020. [23]
1. First shot given
2. 4 weeks later second shot allowed
3. Before 2020 a later second dose
4. 2020 4 weeks after the first
Why it matters

Some children do not respond to the first shot, so the second gives another chance. [23]

How it works

Immunologists call a mother's protection passed to her baby maternal antibodies; it fades within months. [23]
Those antibodies can blunt a shot at six months, so such babies need a third dose. [23]
The truth
- German scientist Catherina Becker says the CDC should advise the earlier second dose. [23]
12
Worriers sleep longer than they say
What happened
- Students prone to worry reported less sleep, but wrist monitors did not record less. [24]
- Villanova University researchers tracked 134 students with wrist monitors for five nights. [24]
- Each step up the five-point scale meant 15 to 20 fewer minutes of reported sleep. [24]
- On average students said 6.74 hours, and the monitors recorded 6.70 hours. [24]
- Sleep worked out from their usual bedtimes came to 7.40 hours. [24]
- The study is in the journal Personality and Individual Differences. [24]
Who is involved

Psychologists call the trait neuroticism: a tendency to feel worry and distress. [24]
- Villanova University one private university
Why it matters

So worried people may feel short of sleep even when the hours are there. [24]

How it works

The authors say worry may shape how people recall and report their sleep. [24]

Students prone to worry said they slept less, while the monitors on their wrists recorded no less sleep.
The idea, in story 01: Why what people report about themselves can differ from what a device records →
The truth
- A monitor can also mistake lying still and awake for sleep, the authors note. [24]
- The students came from one private university, and most were women. [24]
13
Data brokers sell where patients go
What happened
- Phones reveal when people visit doctors, and data brokers sell that, KFF Health News reports. [25]
- The ads offered juice cleanses and Mexican clinics promising recovery with little chemotherapy. [25]
- The US health privacy law, HIPAA, does not cover location data from phones. [25]
- In a 2022 survey, over 92% of patients said health data should not be sold. [25]
- Only 9% of data brokers follow a California law on deleting data, a Stanford study found. [25]
Who is involved

Data brokers are firms that gather people's data, such as phone locations, and sell it. [25]
Chris McQueen, treated for leukaemia, saw ads for fake cancer cures during clinic visits. [25]
- Data brokers gather and sell personal data
- Chris McQueen treated for leukaemia
How it unfolded

California's governor vetoed a bill to block sales of location data on 27 September. [25]

Why it matters

Ads like these can pull sick people from proven treatment toward untested ones. [25]

How it works

A privacy lawyer says a whole trail of locations reveals what one point cannot. [25]
The truth
- Advertising groups say targeted ads keep websites and apps free to use. [25]
- Who gains Data brokers and sellers of unproven cancer cures — Location data from phones at clinics is not covered by US health privacy law, so it can be sold for targeted ads. [25]
- Who pays Cancer patients like Chris McQueen — His phone showed ads for fake cures during visits to his cancer doctor. [25]
14
Fake cosmetics found with mercury and lead
What happened
- Fake beauty products seized in the UK contained mercury, lead and methanol, Trading Standards says. [26]
- Officers tested 50 seized fake cosmetics. [26]
- One skincare product held 13,200 parts per million of mercury. [26]
- One fake perfume was about 80% methanol, a toxic alcohol. [26]
- Fake lipsticks and eyeshadows held high levels of lead. [26]
- About 20% of 18-to-24-year-old beauty buyers knowingly bought fakes last year. [26]
- That was up from 6% in 2023. [26]
Who is involved

Trading Standards is the UK watchdog that investigates fake and unsafe goods. [26]
- Trading Standards UK watchdog
Why it matters

Fakes skip the safety tests and ingredient lists real cosmetics must have. [26]

How it works

Toxicologists warn mercury and lead can harm the kidneys and nerves without visible signs. [26]
The truth
- Nobody can tell from the packaging what chemicals a fake contains, a toxicologist says. [26]
- Who gains Sellers of fake cosmetics — They sell copies of expensive brands without the safety tests real products must pass. [26]
- Who pays Young people buying fake cosmetics — About 20% of 18-to-24-year-old buyers knowingly bought fakes, some holding mercury and lead. [26]
15
Blocking one enzyme helped old mice live longer
What happened
- Blocking one enzyme helped old mice live 17% longer, a study in iScience from China finds. [27]
- AEP grows more active with age and has been linked to Alzheimer's disease. [27]
- Treated mice lived 28 months on average, against 24 months for untreated mice. [27]
- The mice were about 18 months old, middle-to-old age for a mouse, at the start. [27]
- Treated mice also did better on memory, strength and exercise tests. [27]
Who is involved

An enzyme is a protein that does a job; this one, AEP, cuts up other proteins. [27]
How it unfolded
1. 18 months treatment starts
2. 28 months average treated lifespan
Why it matters

The researchers say AEP plays a part in several diseases of ageing at once. [27]

How it works

Biologists test such ideas in mice first; whether they work in people needs separate studies. [27]
The truth
- The mice also ate different amounts, which may explain some of the results. [27]
- AEP has useful jobs too, so long-term safety is still unknown. [27]
- Nobody knows yet
  Whether blocking AEP does anything in people.
  
  The 17% longer life was seen only in mice, which also ate different amounts. [27]
16
Creatine for mood: the evidence is thin
What happened
- Creatine, a supplement known for muscles, is now promoted for mood, but evidence is thin. [28]
- It helps deliver energy to muscles, and small amounts are stored in the brain. [28]
- The brain is 2% of body weight but uses 20% of the body's energy. [28]
- A recent review found 5 trials of creatine for depression, with 238 patients. [28]
- Of the 5 trials, 2 showed a benefit and 3 did not. [28]
Who is involved

The body makes creatine, and meat and fish contain it. [28]
Why it matters

Creatine is one of the most widely used supplements, so claims about mood reach many people. [28]

How it works

In evidence-based medicine, 5 small trials testing different doses cannot add up to a firm answer. [28]
The truth
- Nobody knows the right dose for the brain, or whether that dose is safe. [28]
17
Why some allergy pills cause drowsiness
What happened
- Some allergy pills make people sleepy and slow, and others barely do, an expert explains. [29]
- Older antihistamines date from the 1940s and pass easily from the blood into the brain. [29]
- In one study fexofenadine took up about 2% of the brain's histamine receptors. [29]
- Cetirizine took up 24%, and the older chlorpheniramine 50 to 60%. [29]
Who is involved

There they block histamine, a chemical that also helps keep the brain awake. [29]
How it unfolded

Newer ones like loratadine, cetirizine and fexofenadine arrived in the 1980s and 1990s. [29]
1. 1940s older antihistamines
2. 1980s and 1990s newer ones
Why it matters

Drowsy pills slow reactions, which matters for driving, even when people feel only slightly tired. [29]

How it works

Pharmacologists explain it with the blood-brain barrier, which controls what reaches the brain. [29]
The truth
- Older pills stay on sale partly because people keep buying what they always have. [29]
- Research has not proved these older drugs cause dementia, the article says. [29]
18
A digit in a headline draws more clicks
What happened
- Headlines containing a digit like '3' drew more clicks, a study in Scientific Reports finds. [30]
- At The Washington Post clicks rose 3.6%, and at the website Upworthy 5.2%. [30]
- In a test with 765 people, digits beat both 'several' and the written word 'three'. [30]
- People also remembered more from headlines that carried numerals. [30]
Why it matters

News sites compete for attention, so small cues like digits can steer what people read. [30]

How it works

Cognitive psychologists study attention: what the brain picks out when too much competes for it. [30]
The truth
- The effect was a few percent, measured at two news sites. [30]
19
Judge strikes down a narrow 'frail' rule for Medicaid
What happened
- A federal judge in Massachusetts struck down a narrow US rule on who is 'medically frail'. [31]
- People judged medically frail are exempt from those requirements. [31]
- The rule made frail people prove their condition stops them from working. [31]
- It also listed no conditions that would count as exempt automatically. [31]
- States sued, saying the US federal government had no power to narrow the definition. [31]
- The judge ruled the states had not had enough time to assess who qualifies. [31]
Who is involved

Medicaid is the US government programme that pays for health cover for people on low incomes. [36]
- Medicaid US health cover for people on low incomes
- US states sued
- Federal government wrote the rule
How it unfolded

From 2027, many adults on Medicaid are due to meet work requirements to keep their cover. [31]
1. Thursday the ruling
2. 2027 requirements may slip
Why it matters

A narrower exemption would mean more sick people filing proof to keep their health cover. [31]

How it works

Policy researchers call this administrative burden: paperwork people must do to get help they qualify for. [31]
The truth
- State staff carry a burden too, since they must check each person's status. [31]
- Nobody yet knows when the work requirements will now begin. [31]
What comes next

The ruling could delay the work requirements past the start of 2027. [31]
20
Young Americans swap bars for run clubs
What happened
- Young US adults are joining run clubs and sports leagues to make friends, CNBC reports. [32]
- Volo Sports, which runs adult leagues, expects 950,000 registrations this year, up from 160,000 in 2019. [32]
- Last year it had 710,000. [32]
- In a 2026 poll, about seven in 10 adults under 30 expected to join a league. [32]
- Nearly 24% of New York City Marathon finishers in 2025 were in their 20s. [32]
- In 2019 it was about 15%. [32]
- About 12,000 people paid to run the DJ Diplo's 5km race in New York last week. [32]
- Bank of America found young adults' fitness spending rose faster than bar spending in early 2026. [32]
How it unfolded

The share of Americans who say they drink alcohol is at a record low. [32]

Why it matters

Younger adults have fewer close friends, and a weekly sport gives them a place to meet. [32]
The truth
- Organisers still sell drinks: Volo partners with bars, and Diplo's race had bars too. [32]
- Volo has private-equity investors and has been buying up smaller rivals. [32]
21
Why a face in a dim mirror can turn strange
What happened
- People who stare at their own face in a dim mirror can see it change. [33]
- Around two-thirds reported strong distortions of their faces. [33]
- More than a quarter saw an unfamiliar person; others saw animals or dead relatives. [33]
- Some changes began in under a minute. [33]
Who is involved
- Giovanni Caputo Italian psychologist
How it unfolded

In 2010, Italian psychologist Giovanni Caputo asked 66 volunteers to watch their faces for a minute. [33]
They sat about 40 centimetres from a mirror in a dim room. [33]
In the Bloody Mary dare, children say a name three times into a dark mirror. [33]

Why it matters

That dare sets up the same conditions: dim light, a long stare, and fear. [33]

How it works

Vision scientists call one cause Troxler fading: what stays still at the edge of sight fades. [33]
The brain builds faces from patterns, and in dim light it can build the wrong one. [33]
The truth
- Scientists are still working out exactly how these illusions happen. [33]
22
UAE warns of AI fake images used against children
What happened
- A UAE child safety group warns AI-made fake images are used to bully and blackmail children. [34]
- A child can be targeted without ever sharing a photo of themselves. [34]
- Targets can feel humiliated and anxious, and some become afraid to go to school. [34]
- Some offenders use the images to demand money or more from the child. [34]
- Under UAE law, altering images to harm someone brings at least a year in jail. [34]
- Fines run from Dh250,000 to Dh500,000, about $68,000 to $136,000. [34]
Who is involved

The group is the Child Safety Organisation in Sharjah, part of a family and community council. [34]
- Child Safety Organisation Sharjah, UAE
- The abuser responsible
- The child not to blame
Why it matters

The harm comes from the fake itself, so the child need not have done anything. [34]

How it works

In child protection work, responsibility lies wholly with the abuser, never the child. [34]
The truth
- The group gave no count of how many children have been targeted. [34]
23
Online visits go better with a familiar clinic
What happened
- Japanese patients seen online by a clinic they knew were less often hospitalised, a study finds. [35]
- Researchers studied 189,377 first online visits for respiratory infections from April 2020 to April 2024. [35]
- About 24.7% of visits were with a clinic the patient had used in the past year. [35]
- Of those, 0.4% ended in hospital within 14 days, against 0.9% of the others. [35]
- The study is in Scientific Reports and was funded by the Pfizer Health Research Foundation. [35]
Who is involved

Respiratory infections are illnesses of the airways, such as the throat and lungs. [35]
- Scientific Reports journal
- Pfizer Health Research Foundation funder
How it unfolded

Online doctor visits grew fast in Japan during the Covid pandemic. [35]

Why it matters

Many first online visits are with a doctor who has never seen the patient. [35]

How it works

Doctors call this continuity of care: being seen by the same clinic over time. [35]
The truth
- The study found a link, not proof that knowing the clinic caused the difference. [35]

90% of 5,500 US pupils aged 9 to 17 have used an AI chatbot, and nearly two-thirds use one both for homework and when stressed or lonely, a survey finds

23 stories on this beat today.

Why what people report about themselves can differ from what a device records