You’ve probably seen it on social media before: a paragraph of scrambled text that looks like nonsense at first glance, yet somehow you can read it with surprising ease.

Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoatnt tihng is taht the frist and lsat ltteers be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.

This effect, often playfully referred to as typoglycemia, is frequently shared online as a quirky insight into how our brains work.

But this viral claim is only part of the story. To understand why it works, we need to look at how the brain actually processes written language.

There is no magical ‘rule’

The claim that usually accompanies this snippet is that as long as the first and last letters of a word are in the right place, the order of the middle letters doesn’t matter.

At first glance, the claim seems plausible.

But while there is a kernel of truth here, the explanation is misleading.

Reading scrambled words has much less to do with a magical “rule” about first and last letters, and much more to do with how our brains use context, pattern recognition and prediction.

We don’t read letter by letter

When we read, we typically don’t painstakingly process each letter in sequence. Instead, skilled readers recognise words rapidly by drawing on multiple cues at once. Psycholinguistic research shows that we process words as patterns rather than as sequences of individual sounds.

These include familiar letter patterns, the overall shape of the word and, crucially, the context of the sentence. Our brains are constantly predicting what is likely to come next, then checking those predictions against the visual input.

This is why we often miss typos in our own writing. We don’t see what’s actually on the page, we see what we expect to be there.

The same principle helps us make sense of jumbled words. Even when letters are out of order, enough of the structure remains for the brain to make an educated guess.

Word shape and structure matter

The viral meme suggests that only the first and last letters matter.

But this oversimplifies what’s really going on. We are sensitive to how letters relate to each other within a word. Common spelling patterns and familiar combinations make words easier to recognise, even when slightly distorted.

This is also why certain visual disruptions make reading harder. Text in alternating caps, such as “AlTeRnAtInG CaPs”, is difficult to process because it disrupts the usual visual contour of words. The same goes for “ransom note” lettering made from mismatched fonts, which interferes with pattern recognition.

In other words, readability depends on preserving enough of a word’s internal structure, not just its outer letters.

Not all scrambled text is readable

If the meme were true, any sentence with intact first and last letters should be easy to read. But that’s not what we find.

Take this example:

Salhal I cmorape tehe to a srmmeus day

It follows the supposed “rules”, yet it is much harder to decipher. In fact, this is the opening of Shakespeare’s Sonnet 18: “Shall I compare thee to a summer’s day?”

So why is the viral paragraph so much easier to read? Because it has been carefully (if unconsciously) engineered to be readable.

The hidden tricks behind the meme

Several factors make the famous example easier to process than it appears.

First, many of the words are short, which limits how many possible combinations the letters could form. Words like “you” and “can” are often left unchanged.

Second, function words such as “the”, “and” and “is” are usually intact. These small, common words provide the grammatical scaffolding of the sentence, making it easier to predict what comes next.

Third, when longer words are scrambled, the changes are often minimal. Adjacent letters are swapped (“wrod” for “word”), which is much easier to process than more extreme rearrangements.

Finally, the passage itself is highly predictable. Once you recognise the topic and rhythm, your brain fills in the gaps automatically, much as it does when listening to speech in a noisy environment.

The key to understanding this phenomenon is context. Words are not processed in isolation. Each word is interpreted in relation to the others around it, and within a broader framework of meaning.

This allows us to compensate for missing or distorted information.

But there are limits. As scrambling becomes more extreme, or as words become less predictable, comprehension quickly breaks down. Reading speed also slows noticeably, even when we can still make sense of the text.

Humans and machines

Interestingly, computers can now unscramble jumbled words with remarkable accuracy. By analysing probabilities and patterns across large datasets, algorithms can determine the most likely original form of a word or sentence.

In this sense, machines and humans rely on similar principles. Not rigid rules about letter position, but flexible systems that weigh patterns and probabilities. This highlights why the “typoglycemia” claim is an oversimplification, rather than a scientific rule.

The idea persists because it captures a genuine insight in a catchy way. It reveals that reading is not a simple, letter-by-letter process, but a dynamic interaction between perception and expectation.

At the same time, it’s a reminder of how easily scientific ideas can be distorted as they spread online.

So yes, we can often read scrambled words. But not because the order of letters doesn’t matter. It’s because our brains are remarkably good at making sense of imperfect information. So good, in fact, that they can turn a mess into meaning.

Karen Stollznow does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

Consumption of ultra-processed foods – including soft drinks, snacks and ready meals – is growing worldwide, despite evidence they are unhealthy.

Ultra-processed foods (UPFs) make up about 70% of packaged food products on supermarket shelves, and even more in convenience stores.

In our new research, we explore how companies that produce these foods play on human nature to make such products seem the easiest, most rewarding and compelling option.

We show that UPFs are designed to make us crave them and eat more. They are marketed to all groups, particularly children, in a way that makes them seem the most delicious and convenient option, giving the best value-for-money, despite many health harms.

Our attraction to UPFs is no coincidence. UPF companies combine a range of tactics to drive up consumption. Many of these tactics exploit the ways we think, feel and behave.

Why we keep eating UPFs

UPFs are the most processed foods on the market. According to medical journal The Lancet, they are commercial formulations made from cheap ingredients extracted or derived from whole foods, combined with additives, but mostly containing little to no whole food in the end product.

UPFs are heavily branded and marketed, and most are produced by large international corporations.

But diets high in UPFs carry a risk of developing a wide range of serious health conditions, including excess weight or obesity, type 2 diabetes, hypertension, heart disease, cancer, chronic kidney disease and depression, as well as premature death.

Our research asked why we keep eating diets high in UPFs when we know how unhealthy they are. To answer this, we decided to zoom out and explore the system that develops, produces and markets UPFs, and investigate how human nature is caught up in it.

We reviewed a decade of published research on the food science and marketing of UPFs, and then worked with experts in these fields to create and refine system diagrams to visualise how it works.

These maps are called “causal loop diagrams”, and their power is in showing reinforcing (positive) feedback loops that drive the system towards its ultimate purpose: selling more UPFs.

We found the system is made up of many interconnected loops that capture parts of human behaviour and biology as key elements.

Products designed for maximum consumption

One feedback loop includes the use of addictive combinations of ingredients, particularly refined carbohydrates and fats. Biologically, carbohydrates (including but not limited to sugars) and fats activate different reward pathways between the gut and brain. When they are consumed together, their effects become addictive.

These ingredients can be combined in many different concentrations to hit sensory “sweet spots”. In other words, they maximise pleasure and craving responses while minimising negative responses.

Further strategies include processing methods that suppress peoples’ natural sense of being full or speed up digestion in order to give an immediate but quickly fading sense of “reward”, making us want more, sooner.

UPF marketing strategies

In terms of marketing, products are formulated to be easy and convenient to store and eat, and to appeal to our sense of getting good value.

Various promotional techniques aim to capture consumers’ attention and desire, as well as giving the illusion of healthiness. Strategies targeting children in particular employ popular culture associations with coolness or fun.

Another example of a feedback loop is how corporations collect large and complex data on our purchasing habits and our online lives, informing targeted digital marketing on social media platforms. This tends to be effective at driving purchases, providing more data to further refine these promotion strategies.

Overall we identified 11 different reinforcing feedback loops. Our research is the first study to show this web as part of the UPF system, designed to essentially trap people into buying and eating more and displacing healthier options in diets.

This product-level system also connects with feedback loops further up the supply chain in economic and financial spheres of the global UPF production.

This matters because unhealthy diets and excess body weight cause 18% of preventable premature death and disability in New Zealand. Both risk factors are linked to eating too much UPF.

Unfortunately, New Zealand hasn’t undertaken national nutrition surveys since the 2000s and we have to rely on data from similar countries such as Australia to estimate that UPFs make up about half of our energy intake.

What to do about it

Diets high in UPFs are not the result of people’s free personal choice or weak willpower, but of an intentionally designed system.

Our research shines light on how the UPF system is taking advantage, particularly of children. International experts have framed UPFs as a major global health issue, and advise strong government policy to regulate these products to counter some of these mechanisms.

Policy leadership already exists in other parts of the world, particularly in Latin America. New Zealand could follow other countries that have implemented taxes on UPFs and sugary drinks, regulations restricting advertising to children, strong front-of-pack labelling and transparency policies such as public disclosure of lobbying in government.

Complacency is not an option. The food system needs rebalancing so that it serves and nourishes people now and in the future.

The authors acknowledge the research contribution by Dr Joshua Clark.

Kelly Garton receives funding from the New Zealand Heart Foundation. She is affiliated with the advocacy group Health Coalition Aotearoa.

Boyd Swinburn is affiliated with the Health Coalition Aotearoa.

The Trump administration has called on TV network ABC to “take a stand” after a joke from its late night comedy host Jimmy Kimmel offended the US president and first lady.

Two days before the White House Correpondents’ dinner on April 25, Kimmel broadcast what he said was a “roast” of the Trump administration. Roasts are typically quite savage comedic attacks which have become a traditional part of the dinner.

Trump, who was famously the target of jokes from former president, Barack Obama, at a dinner in 2011, had never attended the dinner while in office. This year he opted to attend, but the comedian’s spot was taken by what was described as a “mentalist”.

So Kimmel said he decided to supply the roast on his show as an “all-American” version of the Correspondents’ Dinner. In what he said was a joke about the 24-year age difference between the couple, he described Melania Trump as “having a glow like an expectant widow”. But after a would-be assassin tried to launch a murderous attack two days later at the dinner, the Trumps have demanded his sacking.

“Enough is enough. It is time for ABC to take a stand. How many times will ABC’s leadership enable Kimmel’s atrocious behaviour at the expense of our community,” Melania Trump wrote in a post on X.

But it appears that ABC, a subsidiary of Disney, is instead standing by Kimmel, who has not been taken off air, in contrast to an episode in September 2025 when Kimmel was suspended after comments he made following the death of right-wing influencer Charlie Kirk, a close friend of the Trumps. After a public outcry, ABC relented and restored Kimmel’s show.

In response, Brendan Carr, the head of the Federal Communications Commission (FCC) has brought forward a review of ABC’s station licences, which were previously not scheduled until 2028 or later. Carr’s actions follow a press conference at the White House on April 26 at which press secretary Karoline Leavitt said coverage critical of Trump, including from his Democrat opponents, was responsible for the rise in political violence in the US by creating what she called a “leftwing cult of hatred”.

These examples highlight the politicisation of “free speech” by the Trump administration as a cudgel to silence disfavoured viewpoints under the guise protecting the public from harm.

First amendment protection for free speech

But these political debates are becoming increasingly distanced from the first amendment. That is, the interpretation of the first amendment by the Supreme Court and the protections it provides to individuals and entities, including media outlets and broadcast companies, from government interference. The wider this gulf becomes, the greater the space between the principles underlying the expansive protections afforded to speech in the US and the public’s understanding of the democratic principles that underpin these protections.

This is more important than ever in the Trump era. Actions taken by the administration to target broadcast networks and individuals for political speech are precisely what the first amendment protects against. It was designed, among other things, to protect individuals, entities and the press from government interference by creating an open marketplace in which ideas compete freely.

This is particularly true for dissenting political speech, which is the core of the first amendment. This explains why government interference with speech based on “the specific motivating ideology or the opinion or perspective of the speaker” – known as “viewpoint discrimination” – is expressly prohibited.

Additionally, whether and to what extent speech is offensive is irrelevant to the protection it enjoys. When it comes to the value of public debate, the first amendment is not neutral. Indeed, as a Supreme Court judgment, Baumgartner v. United States (1944) found: “One of the prerogatives of American citizenship is the right to criticize public men and measures.” Moreover a more recent judgment, Hustler Magazine, Inc. v. Falwell (1988), found that “robust political debate” is expressly encouraged, given that such debate “is bound to produce speech that is critical of those who hold public office”.

Importantly, the Supreme Court found in New York Times Co. v. Sullivan (1964) that such criticism, inevitably, will not always be reasoned or moderate and that public figures as well as public officials will be subjected to “vehement, caustic, and sometimes unpleasantly sharp attacks”.

Read more: New York Times v Sullivan: the 60-year old Supreme Court judgment that press freedom depends on in Trump era

The motive of the speaker is also irrelevant, as the Supreme Court held in Hustler v Falwell that while a “bad motive” may be deemed controlling for tort liability and in other areas of the law, “the first amendment prohibits such a result in the area of public debate about public figures”.

Stakes couldn’t be higher

By expressly linking Democrat criticisms of the president, and pointed critiques (however off-colour) from Kimmel and his fellow political satirists to an upsurge in political violence, the Trump administration is trying to silence criticism of its actions. But it’s also clear that this behaviour is precisely what the first amendment prohibits.

Ironically, the media often portrays these episodes as “feuds” between Trump and his critics.

But when viewed through the lens of the first amendment and its core values in this context, the stakes are much higher. These episodes constitute an effort to wrest control of public discourse by interfering in the marketplace of ideas in order silence those critical of the government.

And history tells us that a government that can silence its critics often does so in pursuit of unchecked power. Viewed through this lens, perhaps the greatest threat to American democracy is the government itself.

Eliza Bechtold does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.

To mark International Dawn Chorus day we’ve asked wildlife experts to make their case for why their favourite songbird deserves your vote. Cast your vote in the poll at the end of the article and let us know why in the comments. We hope their words will inspire you to step outside and soak up some birdsong this spring.

Song thrush

Championed by Cannelle Tassin de Montaigu, Research Fellow in Ecology and Evolution, University of Sussex

When people talk about the UK’s best bird songs they often go straight for the big names – loud, dramatic performers that grab your attention. But quietly in the background is the song thrush, a bird whose song is far more impressive than it first appears.

What sets the song thrush apart is not volume or flair, but structure. Its song is built from short, clear phrases, each repeated two or three times before moving on. It’s as if the bird is politely checking that its audience is paying attention. In a dawn chorus that often feels a bit chaotic, there’s something refreshingly organised about it. It’s a bird that’s actually thought things through.

It might not have the dramatic flair of the common nightingale, and it’s less showy than some of the usual favourites. There are no soaring crescendos or dramatic flourishes. But that’s part of its charm. The song is neat, rhythmic and surprisingly memorable once you start listening for it.

In the early morning soundscape, where many birds seem determined to out-sing one another, the song thrush isn’t trying to steal the spotlight. It just quietly does its thing, and does it very well. Underrated? Definitely. Worth your vote? I’d say so.

Robin

Championed by Judith Lock, Principal Teaching Fellow in Ecology and Evolution University of Southampton

The European robin is a delightfully common sight in gardens. You will very likely have heard the characteristic “tic”, followed by a tuneful verse lasting a few seconds. In noisy urban environments they sing louder, less complex songs, in order to be heard.

The male robins use their spring song (January to June) to signal their quality to females, then forming breeding pairs, and to signal competitive ability to other males. The spring song lasts one to three seconds, composed of four to six short motifs. They have an impressive repertoire of about 1,300 motifs, indicating that song is the particularly important for robins, in comparison to birds that rely more on colourful plumage or behavioural displays to communicate with each other.

Most birds sing mainly in the morning but robins sing all day. People often mistake their lovely evening song for a nightingale’s. Constant territory defence from non-migrating robins means that the robin song is a year-round soundtrack too. From July to December, both males and females sing the autumn song, of higher-pitched long, descending notes, with interspersed warbles. This song is to defend their individual winter territories. This indicates that song first evolved first in songbirds to ensure survival, before it became a signal used by males for reproduction. Each robin’s song is dynamic, constantly changing in response to the condition and age of the bird, and their rival.

Great tit

Championed by Josh Firth, Associate Professor of Behavioural Ecology, University of Leeds

Its song may not be as flashy as the nightingale or as poetically melancholy as the blackbird. But scientists have been taught so much by the great tit’s song, heard across British habitats from ancient woodlands to urban gardens. This spring marks 80 continuous years of UK-based scientists studying great tits at Wytham Woods, Oxford, the world’s longest-running study of individually-marked animals.

The unique dataset includes a family tree totaling over 100,000 great tits, with some birds’ lineages traceable back 37 generations. Early research on Wytham’s great tits during 1970s-1980s resulted in some the first studies to inform the scientific world about how bird song can help males find mates and defend territories, how larger song repertoires can bring more reproductive success, and how young birds learn these repertoires from neighbours (not just their fathers).

And a pioneering study published in 1987 taught us how male great tit song even tracks female fertility, increasing their singing efforts as their female partner’s egg-laying period approaches, and then quietening after she starts laying. Modern technological advances are allowing insight into the hidden meaning embedded in great tits’ songs. In-depth processing of 109,000 recordings of great tit songs has revealed how each bird’s melody tells the story of their own identity as well as that of their local culture and social circles.

A great tit’s age also affects their song: older males keep singing rarer, fading song types while younger birds adopt newer ones. So, Britain’s greatest song belongs to the great tit’s “teacher-teacher” call, for all it has taught us, and for all we have left to learn.

Chaffinch

Championed by Joey Baxter, PhD Candidate in Ecology and Evolutionary Biology, University of Sheffield

Why change a winning formula? As far as I’m concerned, the chaffinch sings the biggest banger that UK birds have to offer. While the blackcap attempts to impress with ostentatious bells and whistles, the chaffinch keeps things simple with a catchy riff. Where the starling goes for quantity and novelty, with a frankly plagiaristic repertoire of mimicry, the chaffinch goes for quality, singing proudly in the knowledge that it is delivering a true earworm.

Bubbling trills accelerate before tumbling downwards, slowing to rich watery chirps and finishing with the final flourish. This jaunty lick, the real hook of the song, is often punctuated by an upward inflection at its end, the rising intonation giving it the air of an unanswered question. The chaffinch’s song has rhythm, it has melody, and it’s instantly recognisable. It possesses the wisdom that sometimes it is better not to do everything, but to do one thing well.

Joey Baxter receives funding from UK Research & Innovation (UKRI), via the Engineering and Physical Sciences Research Council (EPSRC).

Josh Firth receives from UK Research & Innovation (UKRI), via the Natural Environment Research Council (NERC).

Cannelle Tassin de Montaigu and Judith Lock do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

Some medieval texts have barely survived. Beowulf, the Old English masterwork, exists today because of a single manuscript – one that narrowly escaped combustion in 1731. For such texts, the single manuscript is all important. The discovery of another copy would transform our understanding.

By contrast, a work like Bede’s Historia Ecclesiastica Gentis Anglorum (Ecclesiastical History of the English People) survives in more than 160 manuscripts. This volume of material has meant that scholars have tended to focus on just a few of the earliest copies, since these are most likely to preserve a text close to what Bede originally wrote. The result is that many later or less well-known manuscripts have received little detailed attention.

Now, however, computational methods that make it possible to analyse millions of words are changing that picture. Instead of relying on a narrow selection of manuscripts, we can begin to take the full breadth of the tradition into account. And that, in turn, has renewed the value of finding and studying additional copies.

Our own work, motivated by the potential of studying many manuscripts but – for now at least – using traditional methods to locate them, has led to some unexpected discoveries, including, in Rome, a previously overlooked early copy of Bede’s Historia Ecclesiastica. Remarkably, this manuscript also preserves one of the earliest versions of Cædmon’s Hymn, the earliest known poem in English.

Lost and found

The Historia Ecclesiastica was completed in 731 by the Venerable Bede, an English monk often described as the father of English history. It proved to be one of the most influential works of the western Middle Ages. Copies circulated across Europe and the British Isles from the mid-8th to the 16th century.

One of us, Magnanti, was conducting an ongoing hunt for new manuscripts of Bede’s Historia Ecclesiastica, and discovered in the National Central Library in Rome a copy of the text made at the Abbey of Nonantola in the north of Italy, less than a century after Bede’s death in 735. The manuscript had long been presumed lost and, as a result, had never previously been examined in detail by academics.

We have just published details of this discovery in the journal Early Medieval England and its Neighbours.

Rather than being lost, the manuscript had in fact been moved from Nonantola to the church of Santa Croce in Gerusalemme in Rome by the 1650s. During the upheavals of the Napoleonic wars in the early 19th century, it was transferred again to the nearby church of San Bernardo alle Terme, from where it was subsequently stolen, along with other valuable manuscripts.

The book resurfaced in England almost two decades later, when it was acquired by Sir Thomas Phillipps, a 19th-century English book collector and self-described “velomaniac” (manuscript addict). Though Phillips died in 1872, the codex was not sold until 1948, when it entered the collection of the Swiss bibliophile Martin Bodmer. It then disappeared from view once again before being acquired by the National Central Library of Rome via the Austrian-born New York bookseller H.P. Kraus in the 1970s.

Cædmon’s Hymn

The newly rediscovered codex contains perhaps the fifth-oldest surviving complete copy of the Historia Ecclesiastica. As such, it is a hugely important witness to the transmission of Bede’s text to Europe in the century after he completed it.

Even more exciting, the manuscript proved to contain the third-oldest text of Cædmon’s Hymn. Cædmon’s story only survived thanks to Bede. He explains that Cædmon, an agricultural labourer working at Whitby Abbey in north Yorkshire, was at a feast when guests began to recite poems.

Embarrassed that he didn’t know anything suitable, Cædmon left for an early night. A figure then appeared to him in his dreams, telling him to sing about creation, which Cædmon miraculously did, producing his hymn – nine lines of intricately woven praise to God for creating the world.

Cædmon’s Hymn

Translated by Roy M. Liuzza

Now let us praise Heaven-Kingdom’s guardian,

the Maker’s might and his mind’s thoughts,

the work of the glory-father – of every wonder,

eternal Lord. He established a beginning.

He first shaped for men’s sons

Heaven as a roof, the holy Creator;

then middle-earth mankind’s guardian,

eternal Lord, afterwards prepared

the earth for men, the Lord almighty.

While admiring the hymn’s “beauty and dignity”, Bede baulked at including the original English in his Latin. Subsequent readers felt the absence, however, and supplied the original text, in the earliest cases adding it at the end of the Historia Ecclesiastica or in the margin. In the manuscript Magnanti discovered, the hymn appears in the actual text: the earliest such positioning by some 300 years.

Closer examination of the Rome Bede also revealed a major blunder: the scribes appear to have become confused and, between Books I and II of the Historia Ecclesiastica, switched to copying an entirely different text — a sermon on Christ’s descent into hell, prescribed for Easter Sunday preaching. This sermon had passed unrecorded in all the existing catalogues in which the manuscript is described, from 1166 to 2011.

Thanks to computational methods for transcription, collation and textual analysis, a fuller reconstruction of the manuscript tradition of Bede’s Historia Ecclesiastica may now be within reach. That makes discoveries like many the Rome manuscript has yielded just the tip of the iceberg.

The authors do not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and have disclosed no relevant affiliations beyond their academic appointment.

As social media becomes a central part of young people’s lives, concerns are growing about its impact on their mental health. Yet public debates and measures tend to treat adolescents as one homogeneous group. We frequently ignore the fact that social media use does not affect all young people in the same way – nor does it have the same impacts on their wellbeing.

In a recent chapter of the World Happiness Report 2026, published by the UN Sustainable Development Solutions Network in partnership with the University of Oxford, we have examined how problematic social media use relates to the wellbeing of adolescents from different socioeconomic backgrounds.

We looked at 43 countries spanning six broad regions – Anglo-Celtic, Caucasus-Black Sea, Central-Eastern Europe, Mediterranean, Nordic, and Western Europe – covering mainly European countries and their immediate neighbouring areas.

Using data from over 330,000 young people, we found a clear and consistent pattern: higher levels of problematic social media use – that is, compulsive or uncontrolled engagement with social media – are associated with poorer wellbeing.

Teenagers who report more problematic use tend to experience more psychological complaints, such as feeling low, nervous, irritable, or having difficulty sleeping. They also have lower life satisfaction, a measure of how positively they evaluate their lives as a whole.

This pattern appears across all countries in our study, but its strength varies from one country to another. It is particularly pronounced in Anglo-Celtic countries such as the UK and Ireland, while it is comparatively weaker in the Caucasus-Black Sea region.

Socioeconomic background matters

The story does not end with geography. Globally, teenagers from less advantaged backgrounds tend to be more vulnerable to the negative consequences of problematic social media use than their more advantaged peers.

This means socioeconomic status – the material and social resources available to a household, such as income and living conditions – actively shapes the risks and opportunities that young people experience as a result of online environments.

Interestingly, these inequalities are especially visible when we look at life satisfaction. Differences between socioeconomic groups are smaller when it comes to psychological complaints, but much clearer and more consistent for how adolescents evaluate their lives overall.

One likely reason is that life satisfaction is more sensitive to social comparisons. Social media exposes young people to constant benchmarks – what others have, do, and achieve – which can amplify differences in perceived opportunities and resources.

At the same time, these patterns are not identical everywhere. For instance, socioeconomic differences in psychological complaints tend to be modest in most regions including continental European countries such as France, Austria or Belgium, but are more clearly observed in Anglo-Celtic countries such as Scotland and Wales.

In contrast, socioeconomic gaps in life satisfaction appear across most regions, although they tend to be weaker in Mediterranean countries such as Italy, Cyprus and Greece.

A growing problem

We also examined how these patterns have evolved over time. Between 2018 and 2022, the link between problematic social media use and poor adolescent wellbeing became stronger.

This suggests that the risks linked to problematic use may have intensified in recent years, possibly reflecting the growing role of digital technologies in young people’s daily lives, particularly during and after the Covid-19 pandemic.

Importantly, this intensification has affected teenagers across socioeconomic groups in broadly similar ways in most regions. In other words, while inequalities remain they have not widened over this period.

Leer más: Social media addiction disrupts the sleep, moods and social activities of teens and young adults

No one-size-fits-all solution

While public debates about social media and mental health often treat adolescents as a single demographic group, our results show a more complex reality. Problematic social media use is linked to poorer wellbeing across countries, but its effects are shaped by social realities. They vary depending on where young people live and what resources are available to them.

Not all teenagers experience the digital world in the same way, and not all are equally equipped to cope with its pressures. Recognising this is essential for designing policies that are not only effective, but also equitable, ensuring that interventions reach those adolescents who are most vulnerable to digital risks.

A weekly e-mail in English featuring expertise from scholars and researchers. It provides an introduction to the diversity of research coming out of the continent and considers some of the key issues facing European countries. Get the newsletter!

Roger Fernandez-Urbano receives funding from the Spanish Government’s Ministry of Science, Innovation and Universities and the State Research Agency through Ramón y Cajal (RYC) grant. Roger is a member of the International Society for Quality-of-Life Studies (ISQOLS).

Maria Rubio-Cabañez's involvement in this research was supported by the DIGINEQ (Digital Time Use, Adolescent Well-Being and Social Inequalities) project (Grant agreement ID: 101089233), funded by the European Research Council Consolidator Grant.

Pablo Gracia's involvement in this research was supported by the DIGINEQ (Digital Time Use, Adolescent Well-Being and Social Inequalities) project (Grant agreement ID: 101089233), funded by the European Research Council Consolidator Grant.

New research published today in Science Advances reveals that the largest expansion of coral reefs in the past 100 million years happened about 20 to 10 million years ago, between Australia and Southeast Asia.

This vast reef system likely laid the foundations for the extraordinary diversity of marine life we see today.

Coral reefs are among the most diverse ecosystems on Earth. They support about a quarter of all marine species while covering less than 1% of the oceans. Yet scientists have long grappled with the question of how such immense diversity arose in the first place. Where did it begin, and what made it possible?

Our new study uncovers a turning point deep in Earth’s history – a time when reefs didn’t just grow, but expanded on a scale far beyond anything we see today. This expansion may have created the ecological space needed for modern coral reef life to flourish.

An enduring mystery

Biodiversity simply refers to the variety of life in a given place. On coral reefs, this diversity is staggering: thousands of species of fish, corals and other organisms coexist in tightly packed ecosystems.

However, despite decades of research, the origins of this richness have remained an enduring mystery.

Our new study reveals that changes in environmental, biological and tectonic conditions about 20 million years ago promoted the dramatic expansion of coral reefs across a region stretching between Australia and Southeast Asia.

Today, this area is known as the Indo-Australian Archipelago. It’s recognised as a global hotspot of marine biodiversity, especially in an area called the Coral Triangle.

The expansion of reefs in this area coincided with the emergence of many familiar reef organisms, including plating corals and iconic fish groups like parrotfishes.

To uncover this, we combined evidence from geological records, fossils and genetic data. Together, these independent lines of evidence allowed us to pinpoint when and where modern reef biodiversity began to take shape, without relying on any single source alone.

Results suggest reef expansion itself played a crucial role in generating biodiversity. As reefs grew larger, they likely created new habitats and ecological opportunities, allowing species to evolve and diversify.

We have now named this ancient network of reefs the Great Indo-Australian Miocene Reef System. The large reefs in this system were mostly built by corals and crustose coralline algae, an essential group of algae for holding together reef structures. These reefs also provided very important habitat for fish groups that we see on coral reefs today, such as surgeonfishes and butterflyfishes.

Remnants of an epic reef

Surprisingly, the region where this expansion occurred is not where the largest reefs are found today. Instead, reefs off northwestern Australia – including Ashmore Reef, Scott Reef, and the Rowley Shoals – may be remnants of what was once one of the largest reef systems to have ever existed.

Previous geological work has shown this ancient west Australian barrier reef rivalled the extent of the present-day Great Barrier Reef. The new findings go further, suggesting individual reefs within this system may have been far larger than any modern reef.

In fact, the roots of modern marine fish and coral biodiversity may lie in this unexpected place off Australia’s west coast. Over millions of years, biodiversity spread and accumulated elsewhere, particularly across the Indo-Pacific Ocean.

However, there are still uncertainties. Reconstructing ecosystems from millions of years ago requires combining incomplete records. Some aspects of reef size and how these ecosystems connected remain difficult to resolve, as the geological record only contains the remnants of entire reef systems.

But the overall pattern is clear. A massive expansion of reefs about 20 million years ago coincided with the rise of modern marine diversity.

The message is also simple. To understand where biodiversity is today, we need to look deep into the past. The richest ecosystems on Earth may owe their origins to places that no longer appear exceptional – hidden chapters of Earth’s history that continue to shape life in our oceans.

Alexandre Siqueira receives funding from the Australian Research Council through a DECRA Fellowship.

Few things matter more to us than our eyesight. We fear losing it even more than some life-threatening conditions.

Yet for many New Zealanders, access to routine eye care remains out of reach. This is despite the wide-ranging impacts of vision loss for both individuals and society.

It limits opportunities for work and study, raises the risks of traffic accidents and falls, and is linked with higher rates of depression and dementia. Globally, the annual cost in lost productivity has been estimated at nearly NZ$700 billion.

What’s more, it is mostly avoidable. More than 90% of vision loss can be prevented or treated with simple, cost-effective care such as glasses or cataract surgery.

In dollar terms, providing funding for spectacles and eye examinations for New Zealanders could provide a $36 benefit for every $1 spent.

If Aotearoa matched Australia’s public funding policies for community eye care, allocating just 1.2% of its health budget could fund 2.4 million eye examinations and 60,500 pairs of glasses. Current funding delivers eye care services to 25,000 children for about 0.02% of the health budget.

With the government now deliberating its 2026 health budget, our preliminary research looks at what it could cost to make routine eye care a reality for all New Zealanders.

A plight out of sight

Anyone reading this article in New Zealand through a pair of $2 reading glasses isn’t alone in choosing cheap solutions to improve their vision. As many as one in four Kiwi patients may be skipping or delaying specialised eye care because of the cost.

Routine eye examinations and spectacles are delivered almost exclusively by optometrists in private practice, with very little public funding to offset the costs.

This places New Zealand behind other countries, including Australia, the United Kingdom, Ireland and the United States, which fund routine eye care for some or all of their population.

For Kiwis needing financial support for eye care, options are limited. The children of Community Services Card holders can access up to $287.50 for an eye test and glasses via Enable New Zealand.

People on low-incomes can apply for a $280 loan from Work and Income New Zealand, which must be repaid. Spectacles are not currently available in the public sector. Despite advertised “$0 eye tests” and discounted spectacles, the reality is that eye examinations and spectacles remain unaffordable for many.

Optometry services provide more than a new pair of frames. Regular eye examinations are essential to detect and treat progressive conditions such as glaucoma and diabetic retinopathy which are asymptomatic in their early stages.

By excluding this preventative eye care from the public health agenda, New Zealand is leaving some communities to live with an avoidable burden.

In particular, eye care services are two to three times less accessible for Māori and Pacific people than for other New Zealanders.

One recent study found that in an inner-city Auckland community with a high Māori and Pacific population, half of residents with vision loss had never had an eye examination, while three-quarters had never been prescribed custom spectacles.

Should NZ adopt Australia’s model?

If New Zealand seeks a fairer model for eye health, policymakers have only to look across the Tasman.

In Australia, all citizens and permanent residents are eligible for Medicare-funded, comprehensive eye examinations delivered by optometrists.

Around one-third of its population uses these services every year. Uptake is highest among older adults, while additional policies target Indigenous Australians, for example via state-funded spectacle subsidies.

If New Zealand saw similar uptake, we estimate that adopting a comparable model would cost around $349 million a year, funding approximately 2.4 million eye examinations.

An additional $13 million would deliver around 60,500 spectacles to people who need them the most. Even this generous costing is comparable with other health investments, such as the Labour Government’s 2023 proposed investment of $390 million to extend free dental care to approximately 800,000 19–30 year olds.

Universal funding is not the only option: more targeted approaches could prioritise those at greatest risk of avoidable vision loss.

For instance, our analysis indicates that public investment of $89 million could subsidise approximately 760,000 examinations for Community Services Card holders who are most likely to need financial support.

Just $37 million would fund eye care for children under 15 years, aligning with universal dental and GP services for this age group. At the other end of the age spectrum, around $166 million per year would support eye care for older adults, who have the greatest need.

This investment would arguably be more effective than the $61 million proposed within the 2020 health budget to fund one-off “eye health checks”, for which there is no evidence of population-level benefit.

Healthy eyes should not be a luxury. New Zealand can and should include eye examinations and spectacles within its health expenditure.

Preventative eye care is a cost-saving investment that will reduce the societal and economic impacts of vision loss. For policymakers, it as an opportunity to invest in an area of health that has remained out of sight for too long.

Lucy Goodman receives funding from the Health Research Council of New Zealand.

Jacqueline Ramke has received research funding from the Health Research Council of New Zealand, Buchanan Charitable Foundation and the New Zealand Association of Optometrists.

Pushkar Silwal has received funding from Health Research Council of New Zealand.

The US and Israel’s conflict with Iran has drained munitions at an astonishing rate. This is placing pressure on the supply of a crucial metal: tungsten.

Tungsten is used in armour piercing munitions, in components that need to withstand high levels of heat and is an important additive in steel. Militaries around the world would grind to a halt without this strategically important element.

Yet, despite the current demand, the amount of tungsten mined each year is dwarfed by other metal ores, such as iron and aluminium (bauxite).

In addition, most of the world’s tungsten comes from China, which has recently placed restrictions on supplies. For some countries, including the UK, the push to secure new tungsten resources has never been more vital.

The English name for tungsten, comes from the Swedish “tung sten”, meaning heavy stone. Tungsten’s extreme hardness and resistance to thermal shock are what make it sought after for military technology.

In armour piercing munitions, dense tungsten alloy rods use the sheer velocity of their impact to tear through the armour on fighting vehicles and other hardened targets.

When purified, tungsten has the highest melting point of all metals: 3,422°C (6,192°F). Unsurprisingly, it is used in components that need to withstand high temperatures, such as those inside aircraft engines.

Tungsten, along with other metals such as molybdenum, is added to steel to improve its “hot hardness”. Where normal steel would deform at high temperatures, adding the other metals improve steel’s resistance to deformation at high temperatures.

They form carbides with the carbon in steel, making it more resistant to wear, and resisting “creep”, where steel deforms in response to constant stress at high temperatures. Because the tungsten and molybdenum atoms are significantly larger than iron atoms, they improve the “yield strength” of steel, preventing defects in the metal lattice from spreading. Steel is used to make lots of military hardware so tungsten is vitally important.

Limited availability

Having said all that, the global tungsten market is small, tungsten is what is known as a “minor metal”, because it isn’t traded openly on exchanges like the London Metal Exchange. This makes pricing data opaque. While mining operations around the world produce around 2.6 billion tonnes of iron ore every year they only produce around 84,000 tonnes of tungsten.

Tungsten is also considered (alongside tin, tantalum and gold – a group often known as 3TG metals) as a conflict mineral. A significant quantity is mined in regions plagued by violence, forced labour and human rights abuses.

China produces around 80% of global tungsten – and does it so cheaply that it is hard for western firms to compete. In the US, commercial tungsten mining ceased in 2015.

Beijing is leveraging its dominant position to control tungsten supply through a sophisticated state trading and licensing regime. Exports of critical derivatives are restricted to a “whitelist” of authorised state owned firms.

This funnels a huge supply of the metal through a government monitored pipeline. In February 2026, China imposed export controls and reduced mining quotas, limiting tungsten availability. Beijing’s actions have introduced significant friction into western supply chains for military and aerospace applications of tungsten.

Draining stockpiles

Amid the geopolitical turmoil that is unfolding in the Middle East, there is a newfound gargantuan appetite for tungsten, with every bomb, missile and kinetic interceptor further draining stockpiles.

This presents an intractable problem for the defence industry. There has been a 12% increase in the use of military tungsten this year alone – in helicopters, fighter jets and munitions. This is hard to accommodate in a market with no availability.

Global logistics are further complicated by the challenges to global shipping created by the war. This puts a strain on the movement of mining equipment and supplies for processing by the handful of mines outside of China.

Today, there is an economic and strategic opportunity for the UK. The Hemerdon mine in Devon hosts the fourth-largest tin-tungsten deposit in the world, and is a “shovel ready” project being revived by Tungsten West, a mining development company.

Further south, Cornwall Resources’ Redmoor site has revealed high-grade tungsten, tin and copper deposits. This could give the UK a competitive edge in mining and primary extraction, given the current market conditions.

Tungsten also has a recycling rate of 42%, which is higher than for many other critical materials. The recycling rate is the proportion of end-of-life tungsten that is diverted from landfill and made available for reuse. Around 30-35% of the global tungsten supply is derived from scrap (which is to say the proportion of new material made from recycled content).

In western markets, this figure is approximately double – around 70%, because of China’s dominance of the tungsten market. This scrap comes from both manufacturing waste and end-of-life products.

However, supply shortages can often be a catalyst for innovation. In the second world war, metallurgists faced a critical shortage of molybdenum. German U-boat attacks on shipping convoys stymied supplies of this material. This forced metallurgists at UK engineering company Vickers to innovate, and recycle molybdenum from mining drill bits.

In the past, war has forced innovation to ensure the flow of critical materials - We can learn lessons from Britain’s response to molybdenum shortage in the second world war.

The limited global tungsten supply continues to present significant challenges for many countries. One factor that limits stocks is deteriorating ore grades from primary supply (which is to say the concentration of valuable metal inside mined rock is dropping over time). Another is the restrictive export licensing from China.

The current situation has pushed prices to historic highs and challenges the just-in-time nature of many supply chains.

Gavin D. J. Harper receives funding from the EPSRC funded REcycling CRitical Elements in Advanced Technologies for the Environment, RECREATE project (EP/Y53058X/1).

Mehmet Oz, the Centers for Medicare & Medicaid Services administrator, is ordering all states to step up their efforts to crack down on Medicaid fraud.

His April 21, 2026, announcement expanded on the Trump administration’s related enforcement actions, such as withholding Medicaid funds from Minnesota and threatening to do that for New York, California and Maine.

The Trump administration says there’s a big problem with fraud tied to government-funded care delivered in a person’s home or in the community, officially known as home and community-based services, along with nonmedical transportation, behavioral health and new or high billing providers.

The agency Oz leads is now asking states to immediately “revalidate” providers they claim are “high risk.” That is, states are supposed to require providers to prove that they remain eligible to participate in Medicaid and bill the program. The providers primarily offer at-home care, transportation, behavioral health and other services.

Related legislative initiatives, sponsored by Republicans, are also pending in Congress.

We are health services researchers who study the development and growth of the Medicaid home and community-based services program. One of us (Barkoff) previously served in the role of administrator and assistant secretary for aging of the Administration for Community Living.

We strongly believe that it is sensible and necessary for the government to take steps to prevent, root out and punish Medicaid fraud. But we are mindful that the government’s anti-fraud strategies and tactics could unnecessarily disrupt the very services that people depend on day to day.

We aren’t alone. Many researchers, advocates and policy experts are alarmed by this White House policy.

What’s at stake

All told, Medicaid provides health insurance coverage for about 75 million low-income Americans, including many who are at least 65 years old.

More than 5 million Americans benefit from government-funded home care, which is aimed at keeping low-income people with disabilities and frail older people living in their own homes and communities. Medicaid pays for most home care, covering nearly two-thirds of all such spending in 2023.

When home care works well, people become less likely to have to move into nursing homes or other assisted living facilities. If a home care aide doesn’t show up, the consequences are immediate and can be dire.

An older adult may be unable to get out of bed. A person with disabilities may miss meals or medications. A family caregiver may have to take time off without notifying their employer in advance and lose wages.

That’s why the federal government’s actions – particularly those targeting services provided in a person’s home or community – could endanger millions of people.

Supreme Court ruling

Home and community-based services include help with bathing, dressing, eating, medication management and mobility – services that allow people to remain in their homes rather than moving into nursing facilities. A shift toward home-based care has been underway for decades, driven by both costs and civil rights protections.

One reason for the shift was the 6-3 ruling in 1999 by the Supreme Court in the Olmstead v. L.C. case. The majority affirmed the right of people with disabilities to live in their own homes and communities when possible.

Today, most Medicaid long-term care spending covers the cost of services provided in a person’s home or local community, rather than in an institution. These services cost less and lead to better outcomes. Research and program data consistently show that fraud in these programs is relatively rare.

This is especially true due to safeguards like electronic visit verification, which ensures providers are actually providing services in the home.

Another safeguard in place is that most states contract with and approve fiscal intermediaries, which act as payroll, payment and compliance managers, to make sure that there are verifiable records, payment controls and audit trails in place for the Medicaid program.

Not all ‘improper payments’ are fraudulent

A central problem in the Trump administration’s strategy to root out alleged Medicaid fraud is a basic mischaracterization of many things as fraud that aren’t fraudulent.

Federal agencies are tracking “improper payments” and incorrectly equating them with fraud. The Centers for Medicare & Medicaid Services makes clear that most improper payments stem from documentation errors or administrative issues – not intentional wrongdoing.

The Government Accountability Office, a nonpartisan government agency that produces in-depth research, notes that while fraud does lead to improper payments, the reverse is not necessarily true. That is, improper payments can have a cause besides fraud, such as administrative errors and eligibility processing mistakes.

Blurring the distinction between improper payments and fraud can make it seem like providers are illegally taking advantage of the system. And when that happens, policymakers may turn to blunt solutions that do little to punish actual fraudsters, such as cutting or withholding funding, rather than fixing administrative problems.

Exaggerating the scope of Medicaid fraud

In our view, proposals to overhaul Medicaid’s enforcement methods should be grounded in strong and objective data. Yet much of the argument for structural reform relies on anecdotal examples, isolated cases and select audit findings without broader context.

One of the more egregious cases perpetrated by providers was a home care agency in Pennsylvania that billed fraudulent claims between 2020-2023 totaling US$1.8 million. Another was the Minnesota provider penalized in 2025 after billing for services not delivered.

But those infractions do not justify characterizing an entire category of services that helps tens of millions of Americans remain in their homes as rife with fraud and in need of dramatic changes.

By contrast, as we explained in Health Affairs ForeFront in March 2026, federal oversight bodies – including the Centers for Medicare & Medicaid Services, the Government Accountability Office and the Health and Human Services’ inspector general – produce systematic, data-driven analyses.

These sources consistently caution against equating improper payments with fraud and emphasize targeted approaches to program integrity.

The government designated some 6% of the annual payments by Medicaid as improper payments between 2022-2025, which were worth about $37 billion. Yet, more than 3 in 4 improper payments resulted from insufficient documentation, which usually doesn’t indicate fraud or abuse.

What’s more, Medicaid fraud is regularly subject to enforcement actions. In 2025, Medicaid fraud control units reported 1,185 convictions for fraud nationwide, and combined recoveries from criminal and civil cases totaled about $2 billion.

Emphasis on high-profile cases

Again, a few widely publicized fraud cases in Minnesota and a few other states do not prove that fraud is a chronic problem for Medicaid billing in home care programs.

In an agency as big as Medicaid, which spends nearly $1 trillion annually, some level of fraud will occur. The key question is whether fraud is widespread, systemic or goes unpunished.

Available evidence suggests Medicaid fraud is none of those things.

For example, large-scale home care programs serving hundreds of thousands of people report extremely low rates of confirmed fraud cases. Enforcement data from Medicaid Fraud Control Units, which investigate and prosecute Medicaid fraud, show that when fraud occurs, it is investigated and prosecuted.

In other words, the presence of enforcement activity is evidence that oversight systems are working – not that they’re failing.

A better approach

We believe that many strategies that are better than those that the Trump administration is embracing are readily available. Some examples include improved data analytics, stronger referral systems within managed care plans, enhanced provider screening and documentation standards, and continued support for Medicaid Fraud Control Units.

These approaches target fraud directly without jeopardizing access to the essential Medicaid services that help tens of millions of older adults and disabled people remain where they want to be: in their own homes instead of in more expensive nursing homes.

Given that the U.S. spends about $930 billion a year on the program, we don’t question the wisdom of engaging in its oversight.

But we are concerned that the policy response to alleged fraud could harm the very people that the Trump administration says its efforts are meant to protect.

Marc Cohen receives funding from the RRF Foundation for Aging.

Alison Barkoff receives funding from the Commonwealth Fund.

Jane Tavares receives funding from the RRF Foundation for Aging.

Sara Rosenbaum receives funding from the Commonwealth Fund.

If you suffer from information overload, or are unsure what to trust online, you’re not alone. Australians are increasingly disengaging from traditional news, turning instead to social media, influencers and – more recently – generative artificial intelligence (AI) chatbots and summaries.

It’s a murky, polluted world where opaque algorithms decide what you see. They’re known to have little regard for accuracy, quality or the evidence-based reporting we need for a safe and thriving community.

At the same time, local journalism is disappearing. Distrust in mainstream news is growing. This issue has escalated rapidly with “zero-click” AI search results. Instead of serving links, they show the information upfront. This decreases traffic to news websites, further reducing audience, subscription opportunities and revenue. The rapid spread of AI has pushed an already fragile news ecosystem closer to breaking point.

Earlier this year, a News Futures: Media Policy Roundtable brought together 45 leaders from industry, government, not-for-profit organisations, digital platforms and academia.

The attendees agreed that the opacity of algorithms on social media, search and AI platforms – which decide what is shown, ranked or omitted with little accountability – has become a core threat to journalism and audience trust. Published today, the resulting report proposes a paradigm shift in how we support and define journalism in Australia.

Misinformation is flourishing

Misinformation flourishes when there is high demand for information but insufficient verified evidence. A healthy (and prominent) supply of quality news and information can counterbalance misinformation. Our research shows a strong link between news consumption and people’s ability to verify misinformation.

For consumers, laws and civic education have not kept pace with AI content, such as deepfakes. There are no clear standards for showing where online content comes from or standard guidelines for checking if it’s real. Because many AI systems work like black boxes, it’s also hard to know who is responsible when they make mistakes or show bias.

Australians already have very low confidence in their ability to verify misinformation. Only about 40% are confident they can check if a website or social media post can be trusted, and only 43% are confident they can check if information they find online is true.

This problem is being compounded by the growing prevalence of AI slop and hallucinations (low-quality and erroneous content). In fact, Australians are among the most concerned about online misinformation globally.

Read more: Slopaganda wars: how (and why) the US and Iran are flooding the zone with viral AI-generated noise

People don’t know whom to trust

Experts at the roundtable were worried about low media and AI literacy among citizens. Many Australians struggle to verify information online, and are unsure where to turn for trusted sources.

When everything starts to look unreliable, switching off can feel like the safest option, which many Australians choose to do – 69% avoid news often, sometimes or occasionally.

The problem is digital platforms are an unreliable interface for news. Through algorithms, they make invisible and unaccountable choices that reshape the public’s access to information. In selecting where information is drawn from, these digital intermediaries can create new “winners” and “losers”, elevating some content above others with little regard for quality or accuracy.

But there is no impetus for platforms to explain how their algorithms work or when they change, how news is prioritised (or de-prioritised), or how AI-generated information is produced.

There is an urgent need for transparency in algorithmic curation and mandatory labelling of AI-generated content.

Where to from here?

The roundtable participants identified five priorities that, together, could drastically improve our information ecosystem. Three of those specifically target AI.

1. Greater transparency from big tech platforms. Australians deserve to know how algorithms curate news on search engines, social media, and AI chatbots. They also need to know when AI is involved in producing content. Clear labelling and disclosure rules would help rebuild trust and give users more control.

2. Fair rules for AI use of news. AI companies should not be able to take journalism for free. Industry-wide licensing agreements, copyright reform and stronger competition law could ensure news organisations are compensated when their work is used to train generative AI tools.

3. Prioritising media and AI literacy education across the nation. Educating people on how algorithms work, and how to spot bias and misinformation is one of the fastest and most cost-effective interventions available. And it’s not just for schools – adults need ongoing opportunities to upskill too.

4. Journalism funding should reflect its role as a public good. One-off grants are not enough. Proposals such as a tax offset for journalists’ salaries is a sustainable alternative that could support newsrooms directly, especially small and regional outlets, while remaining accountable.

5. Journalism training for news influencers, content creators and digital-first outlets. A common industry code is required to ensure the quality of the whole news ecosystem, and the industry needs to work on this together.

Society can’t afford an information environment in which invisible AI dictates what we see. Without action, the public interest journalism that underpins democracy and social cohesion will continue to crumble.

Sora Park receives funding from the Australian Research Council, Creative Australia and the Department of Infrastructure, Transportation, Regional Development, Communications, Sport & the Arts.

Janet Fulton receives funding from the Department of Infrastructure, Transportation, Regional Development, Communications, Sport & the Arts.

Saffron Howden receives funding from the Commonwealth through an Australian Government Research Training Program Scholarship.

Momoko Fujita does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.