An autonomous agent powered by OpenAI’s advanced artificial intelligence (AI) models went rogue during a security test and hacked multi-billion dollar tech startup, Hugging Face, last week.

The agent didn’t just exploit vulnerabilities in Hugging Face’s systems to achieve what it perceived as a strategic gain. It also exploited vulnerabilities within OpenAI’s infrastructure.

Of course, hacks are very common cyber threats that organisations face frequently. But this incident is different, because the AI agent acted without any human input. It signals a seismic shift in cybersecurity, and shows that governments and tech companies need to take urgent action to prevent this risk escalating.

Even OpenAI described the attack as “unprecedented” and acknowledged it expects similar ones “to become more commonplace with the proliferation of increasingly cyber-capable models”.

A company under attack

Hugging Face is famous in the AI space. Its mission is to “democratise good machine learning” by providing benchmark datasets, community collaboration tools, and robotic platforms. The company is valued at US$4.5 billion.

On July 16, the company announced it had been attacked, with a hacker obtaining unauthorised access to some internal datasets and credentials. It said the hacker was likely “an autonomous AI agent system” due to the sophistication of the attack.

Five days later, Open AI announced the attack had been driven by some of its models: GPT-5.6 Sol and a yet-to-be released model.

The tech giant was conducting what are known as “red teaming” exercises. These are essentially simulated cyber attacks that help identify the capabilities, risks and vulnerabilities of AI systems before they are publicly released. They are typically conducted within an isolated environment to ensure potentially dangerous systems do not escape and cause harm to real systems.

But in this case, the AI agent did escape – even though OpenAI had some guardrails in place to prevent this.

Hugging Face became a lucrative opportunity for the AI agent. It hosts ExploitGym, a benchmark that tests an AI agent’s ability to exploit real-world systems. The AI decided to turn every stone upside down to obtain access. With persistence, it succeeded.

Hugging Face was confronted with a challenge when attempting to use external AI services to diagnose the problem. The guardrails around more advanced models such as GPT-5.6 Sol and Claude Fable 5 are intended to stop them being used for cyber attacks – but they can also stop the models being used for sophisticated cyber defence.

So Hugging Face resorted to using an open-source model, GLM5.2, developed by the Chinese company Z.AI, to counter the cyber attack.

Hugging Face said GLM5.2 was an advantage because it was not exposed to the attack data. Both Hugging Face and Open AI are collaborating on forensic analysis, post-incident recovery and risk mitigation strategies.

More sophisticated threats are coming

A March 2025 study by the United Kingdom’s AI Security Institute showed the best AI could complete 80% of the steps needed to gain full control of a portion of an external system. Within four months, it reached 100%.

Z.AI’s GLM5.2 was only released in June, with 744 billion internal variables, known in the world of AI as “parameters”. The fact that Hugging Face assessed, vetted and deployed it within four weeks should be an eye-opener for organisations with long acquisition cycles.

The connectivity we all enjoy today can equally be our greatest threat. Cyber threats spread faster than human viruses and can create economic damage similar in magnitude to a country’s GDP.

More sophisticated cyber threats – the kind exemplified by the Hugging Face hack – will exploit the security layers that humans designed for human attackers, regardless of how sophisticated our designs are.

Indeed, in this particular case, even OpenAI’s own understanding of its models couldn’t predict or contain the rogue AI agent. This shows the need for all AI companies to urgently update and strengthen their guardrails, in order to help prevent a similar attack occurring with far more devastating consequences.

It is good to see Hugging Face and OpenAI collaborating on the investigation into the attack. This showcases the importance of putting aside market competition and blame when the situation demands.

An early warning

The fact that Hugging Face used Z.AI’s open-source model to diagnose and counter the attack also shows the advantages of not relying on just a few pieces of tech.

States that are not in the game of developing their own AI models need to learn from this incident the value of being different. It is not too late to design new models that could save us in situations when the most advanced models fail – or, even worse, attack us.

Indeed, last week, another Chinese company, Moonshot AI, released Kimi K3. This model has 2.8 trillion parameters, its advanced performance stunning the tech world.

It is no longer a question of “if” AI agents go rogue and attack us by themselves. The Hugging Face incident is an early warning that we must accelerate our preparedness. The threat is real and here.

Hussein Abbass, Professor, School of Systems and Computing, UNSW

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The Victorian government has announced it will introduce “nation-leading” laws to make technology companies accountable for preventing hate and keeping children safe.

The announcement comes as governments around the world grapple with how best to mitigate online harms, including emerging ones associated with artificial intelligence (AI).

So what are the new laws? And how will they work in practice?

Unmasking trolls

The state will introduce a new “unmasking” law. This will enable Victorians who are being vilified online to find out who is targeting them via anonymous accounts.

The law will give the Victorian Civil and Administrative Tribunal (VCAT) the ability to issue a “demasking order” to force social media platforms to identify anonymous users.

“Hiding behind an anonymous account online should never mean you can vilify someone and get away with it,” Victorian Premier Jacinta Allen said in announcing the new law.

This new law follows the April 2026 introduction of stronger civil protections for people targeted with vilification because of a protected attribute, such as their race, religion, disability, gender identity, sex, sex characteristics or sexual orientation.

It also comes amid growing debate about when someone should be afforded anonymity offline.

As antisemitism adviser and former chief scientist Alan Finkel told the Royal Commission on Antisemitism and Social Cohesion last week:

I do feel that the cloak of secrecy across our community, not just in universities, is possibly overused, especially if somebody has been found to have breached a code.

Making it easier to sue

Victoria will also introduce a new law to make it easier for families to sue social media platforms and AI providers for negligence causing psychiatric harm to a child.

Currently, there is a legal threshold requiring “proof of a permanent impairment of 10% or more” before a family can bring an action before the courts. The new law will remove that threshhold in cases brought on behalf of children.

In making this announcement, the government pointed to overseas courts that have already found platforms liable for designing addictive features aimed at children.

For example, in March 2026, a court in the United States found Meta and Google were negligent in the design of their platforms, Instagram and YouTube. The companies were ordered to pay millions to a young woman who argued the social media platforms exacerbated her mental health conditions.

The government said it will consider whether to extend similar protections to adults.

Several points of concern

These laws are not yet drafted, let alone implemented. So how they will work is not yet clear. The government’s announcement explains these laws “will be developed following targeted consultation with VCAT, the courts and other stakeholders”.

The time needed for consultation and to draft the laws may also prove challenging, with a state election set to be held in November. That leaves only a few weeks of sitting time for the government to pass these reforms ahead of the election.

The consultation period is also likely to raise several points of concern.

Some cyber security experts have criticised the proposal. For example, Susan McLean, a former Victorian police officer who now specialises in cyber safety, said it “highlights the Victorian government’s lack of understanding of the digital world”.

One key criticism is that the courts can already compel technology companies to identify anonymous users, with police able to access this information through social media platforms’ law enforcement portals.

But as Susan McLean acknowledged, many cases “go unreported [as] people don’t make a police report in the first place”.

While extending VCAT’s powers may facilitate the process for identifying anonymous users in civil cases, this still places the burden on consumers to sue technology companies.

This means the benefits of this new law may be beyond the reach of those who cannot afford to pursue litigation.

Another concern is how the law would be applied to users in other countries, or those who pretend to be outside of Victoria by using VPNs. It is also unclear whether technology companies would comply with the laws and what the consequences would be for non-compliance.

Is there a better alternative?

Even if these new laws are passed, consumers living in other states would not benefit from these reforms. Other states would need to introduce similar laws, or the federal government would need to pursue legislation at a federal level.

Victoria’s consultation process may open such discussions, as happened with Australia’s social media ban. That legislation was first put forward by South Australia. It later became national law when it was taken up at a federal level. The same could happen here.

Or it may be that Victoria’s proposed laws become unnecessary as the federal government has committed to introducing digital duty of care legislation.

This law would go further than current social media restrictions to hold technology companies to account for harmful content and design features. That could address both vilification and mental health concerns.

Digital duty of care legislation could protect Australians in every state, without the need for individuals and families to pursue costly and time-consuming litigation.

Whether or not Victoria’s laws are passed, they point to the need for additional regulations to make social media, AI, and other online spaces as safe as they can be for all Australians.

Lisa M. Given, Professor of Information Sciences & Director, Social Change Enabling Impact Platform, RMIT University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Seeing a psychiatrist via telehealth became easier from January 2022. That’s when Medicare-reimbursed telepsychiatry arrangements became permanent.

The idea was for these video and phone consultations to provide more flexible and sooner appointment times, without needing to travel.

But our new analysis shows what happened next.

We show how telepsychiatry became dominated by consultations for stimulant prescriptions since 2022. These were mostly for people with attention-deficit hyperactivity disorder or ADHD.

Telepsychiatry for ADHD specifically did increase access for people living in rural and remote areas. But it was more likely to be provided by psychiatrists living in more affluent major towns and cities.

These types of ADHD telepsychiatry appointments are also associated with greater out-of-pocket costs than the equivalent face-to-face appointments.

A three-fold jump in ADHD scripts

We examined nearly 4 million Medicare Benefits Schedule records of psychiatric consultations between 2017 and 2023. We also looked at the relevant Pharmaceutical Benefits Schedule prescription records arising from these consults.

Before COVID, our study showed less than one in 20 psychiatry consultations were linked to an ADHD prescription, and less than one in 50 were conducted by video.

By 2022, the number of ADHD-associated prescriptions increased almost three-fold from 4.3% to 11.8%. About one in four video psychiatry consults were linked to stimulant prescriptions.

Prescriptions for other psychiatric conditions linked to video consultations barely shifted over the same period, and in some cases fell.

This spike in ADHD scripts via telehealth isn’t simply because more people are being diagnosed with ADHD overall, although that’s also true. Australian Institute of Health and Welfare figures show ADHD medication dispensing rose roughly eleven-fold between 2004–05 and 2023–24.

We found the growth in psychiatric appointments created by telehealth was largely taken up with ADHD assessments. In particular, these are the kind of appointments that end with a new prescription rather than ongoing follow-up care.

Who is prescribing and where?

One of our earlier papers shows how telepsychiatry helps alleviate psychiatrist shortages in rural areas.

So we initially expected ADHD prescribing via telepsychiatry to be concentrated among psychiatrists working in under-served, regional and remote areas – the patients telehealth was explicitly designed to reach.

However, the steepest rise in ADHD prescribing was provided by psychiatrists practising in Australia’s most socioeconomically advantaged postcodes. Almost half of all psychiatrists in our dataset were based in the wealthiest areas. This group accounted for the bulk of the increase in video ADHD prescriptions.

That matters because video ADHD consultations, on average, cost patients more out-of-pocket than an equivalent face-to-face appointment.

So the picture is less about telepsychiatry improving access and removing the cost barrier to care. It’s more about creating well-paid specialist services for people who can already afford it.

Why ADHD, specifically?

Interest in ADHD, and its diagnosis, has exploded for a number of reasons.

Social media has allowed more people to recognise their ADHD-like symptoms, particularly adults who were never assessed as children.

Search terms such as “ADHD treatment” and “ADHD medication” climbed steadily across the study period. We found this online interest tracked closely with the rise in video prescribing of stimulants.

Others point to social media platforms, digital notifications, emails and algorithmic content taxing the brain, making it harder to focus for long periods. This “attention economy”, they argue, may also contribute to people thinking they have ADHD and seeking a diagnosis.

When services for ADHD dominate, others will miss out

None of this means the demand isn’t real, or the diagnoses are wrong. Drug treatment for ADHD has a strong evidence base and can be life-changing.

This shift was strongly associated with a higher proportion of female patients, whose ADHD has been historically under-recognised.

But these video ADHD consultations can be lucrative. They offer a way for psychiatrists to see more patients, faster. So a single “one-off” assessment of 30 minutes or longer that results in a script – the type we identified – is an attractive business model.

When ADHD comes to dominate Medicare funding for telehealth originally designed to broaden access to all psychiatric care, market access for people with other mental illnesses may be crowded out.

So yes, telepsychiatry is expanding access. But our study shows its model is focused on a single, lucrative, fast-growing diagnosis and prescription business.

What does this mean as GPs prepare to step in?

Until recently, only psychiatrists could diagnose and start ADHD medication for adults in most of Australia.

But in response to the difficulty accessing psychiatric consultations, state governments are moving to let GPs diagnose and prescribe for ADHD.

While this sounds sensible, we’d be concerned if similar issues we’ve flagged in telepsychiatry were seen in general practice. That is, an increase in commoditisation of ADHD care.

Luke Sy-Cherng Woon, Visiting Fellow, National Centre for Health Workforce Studies, Australian National University; Jeffrey C.L. Looi, Associate Professor, Social Psychiatry and Epidemiology Research Unit, School of Medicine and Psychology, Australian National University; Tarun Bastiampillai, Professor of Psychiatry, Flinders University, and Wai-Man Liu, Associate Professor, Research School of Finance, Actuarial Studies and Statistics, Australian National University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Earth’s orbit is getting crowded.

About 16,000 satellites currently circle our planet, supporting everything from GPS navigation and weather forecasting to banking, emergency services and internet communications.

Dozens more are launched every few weeks. Some estimates suggest the total number of satellites could exceed 100,000 within this decade, with more conservative estimates landing on up to 60,000 satellites by 2030 – still a staggering amount.

This rapid growth is creating an important challenge. How do we safely manage an increasingly crowded orbital environment while ensuring the satellites we depend on continue to work reliably?

The risks are not difficult to imagine. Large satellite constellations increase light pollution and other disruptions to astronomy and the night sky. More satellites mean more traffic, a greater chance of collisions and an increasing amount of space debris.

In a worst-case scenario, space debris can cause a runaway chain reaction known as Kessler syndrome, which would ensconce Earth in a cloud of debris and render its orbit unusable, without the ability to launch satellites or any other space missions.

Even short of this, ageing or damaged satellites can become hazards if they stop working, collide with other objects, or eventually make uncontrolled re-entries through the atmosphere.

This raises a practical question – satellites can’t simply be brought home for repairs. So how do we maintain tens of thousands of machines that are hundreds of kilometres above Earth?

Satellites don’t last forever

The challenge of satellite maintenance became more visible in March this year when a large NASA satellite made an uncontrolled re-entry into Earth’s atmosphere.

The US Space Force confirmed the spacecraft re-entered over the eastern Pacific Ocean, and NASA expected most of it to burn up, though some components may have survived. The event attracted worldwide attention as experts tracked its descent and estimated where debris might land, including the possibility that large debris could one day cause damage in populated areas.

The incident was a reminder that satellites don’t last forever. Like any machine, they age. Batteries degrade, electronic components wear out and harsh space conditions gradually take their toll.

Unlike aircraft or cars, however, we can’t easily take satellites to a repair workshop.

Once launched, they must continue operating in an environment of intense radiation, extreme temperature changes and constant mechanical stress. Servicing missions are technically possible, but remain expensive and relatively uncommon.

How do we keep satellites ‘healthy’?

Today, satellite health is monitored largely from the ground.

Engineers receive streams of telemetry data showing battery performance, temperatures, power consumption and the status of onboard systems. They analyse this information and look for warning signs that something may be going wrong.

This approach has worked well for decades. But it may become increasingly difficult as satellite constellations grow from dozens of spacecraft to hundreds or even thousands. Human operators can only monitor so much information at once.

This is where recent advances in artificial intelligence (AI) may help. Researchers have been investigating how AI can identify early signs of satellite degradation before they become mission-threatening failures.

One important example involves batteries. Satellite batteries gradually lose performance over time, much like the battery in a smartphone or electric vehicle.

If this degradation can be detected early, operators may be able to adjust how a satellite is used, extend its operational life or avoid unexpected failures. They could do this by sending new instructions to the satellite, such as reducing power-hungry activities, changing when data are processed or transmitted, or placing non-essential systems into standby.

Our recent research used publicly available NASA satellite battery data to explore how machine learning (a type of artificial intelligence) can recognise patterns associated with battery ageing and predict future performance.

The goal is similar to predictive maintenance systems already used in modern aircraft, wind farms and manufacturing plants. Rather than waiting for equipment to fail, AI looks for subtle changes that suggest problems may be developing.

Satellites can learn from each other

In our approach, we also considered federated learning.

Normally, enormous amounts of satellite data would need to be transmitted back to Earth for analysis. This requires time, bandwidth and energy. Federated learning offers a different approach. Individual satellites can “learn” from their own experience and share useful insights with other satellites or ground systems without constantly sending every piece of raw data.

In simple terms, satellites could help each other become better at recognising potential faults. Over time, this could support continuous self-monitoring across large satellite networks.

There are, however, important limitations.

AI can’t prevent every satellite failure. It can’t eliminate space debris or solve orbital congestion on its own. Predictive models require extensive testing, such as checking them against historical satellite data, simulated faults and laboratory battery experiments before they are trusted in orbit. And any autonomous decision-making systems must be reliable enough for safety-critical applications while remaining under human oversight.

The next great challenge of the new space age may not simply be launching another 100,000 satellites. It may be ensuring those satellites are intelligent enough to monitor their own condition, detect problems early and help keep the space services we rely on running safely and reliably.

Tony Jan, Professor of Information Technology and Director of Artificial Intelligence Research and Optimization (AIRO) Centre, Torrens University Australia

This article is republished from The Conversation under a Creative Commons license. Read the original article.

It happens just when you need to leave. Bags are packed. Shoes are on. Then your child decides a jumper is impossible.

You say, “put your jumper on”.

They say, “I’m not cold”.

Do kids really not feel the cold like adults do? Or are they just expressing their independence? And when should you insist?

A clue from pain research

I study how children experience pain, and pain research offers one clue about this jumper battle.

A child may scrape their knee during a game and barely notice until the game stops. The scrape was there throughout, but chasing a friend or reaching the next base kept winning their attention.

Cold can slip into the background in a similar way. A child’s fingers may be cooling while the playground remains far more compelling. Then the game ends.

Their attention returns to their hands at around the same time their moving muscles stop producing so much heat. Suddenly, the jumper may seem like a better idea.

So “I’m not cold” can mean, “I can feel it and I’m comfortable”.

It can also mean, “the jumper feels worse” or “I want to keep playing”. Sometimes it means, “I am four years old and this has become a matter of principle”.

Similarly, when parents ask, “but aren’t you cold?” we’re often asking several questions at once.

Will you still be warm when you stop running? What if the wind picks up? How about later, when we’re standing still on the sideline at the soccer field?

Children report the present moment, and parents factor in the forecast.

What cold actually feels like

Your skin contains sensory nerve endings that respond as its temperature changes.

One cool-sensitive channel, called TRPM8, helps convert cooling at the skin into electrical activity in sensory nerves.

This is the same channel that menthol activates, which is why mint can make your mouth feel cool even when there’s no real temperature change.

And as you know with mint, a strong cooling sensation can sometimes become uncomfortable or even painful. Other factors such as wind, wetness, contact with cold surfaces, movement, and how much time we’re outside can all influence how we experience temperature.

For example, a parent who is standing still in a playground, clutching a coffee, may be acutely aware of the gap in their coat where the icy wind is sneaking in.

But children tend to run, climb and jump in bursts – and moving muscles produce heat.

Children also differ from adults in body size, body composition, metabolism and how their circulation responds to cold.

One 2024 laboratory study, done indoors, looked at children aged six to nine. It found their sedentary metabolic rates (how much energy you’re burning when you’re resting) were around 39% higher compared to adults in the study.

Their skin was also warmer over parts of their torso, and the skin on their hands recovered temperature and bloodflow faster than adults after being exposed to the cold.

So it’s not that kids don’t feel cold at all, but they may have a quite different experience from an adult standing in the same air.

Bodies prepare for what comes next

We often learn about thermoregulation – how the body maintains its core temperature – as though the body were simply a thermostat. The body detects a temperature change, then bloodflow changes, and sweating or shivering bring it back towards the middle.

But our movement and behaviour also play an important role in maintaining this balance. When we’re cold we may walk into sunshine, curl up, or add a layer; when we’re hot we take one off. A child who keeps running may already be generating the warmth they need.

Bodies also prepare for expected demands. Researchers use the term allostasis to describe this wider process of how the body maintains stability through change.

Some adjustments happen automatically. For example, before we exercise, our heart rate and breathing begin adjusting for the work ahead. Others involve choices, such as moving into the sun, seeking shelter or reaching for warmer clothing.

But young children outsource some of this forecasting to adults.

Kids supply the live report from inside their body. We add the weather forecast and the schedule. We pack snacks for hunger that has yet to arrive, spare clothes for puddles yet to be found, and jumpers too.

What the jumper itself feels like

A jumper creates its own sensations. It may feel scratchy, bulky or restrictive. It can make climbing harder, then become hot and sweaty as soon as the child starts running.

Tags, seams and some fabrics can feel intensely unpleasant, especially for children with tactile sensitivities. Clothing tags and light touch, for example, can cause marked discomfort for some autistic children.

A child may genuinely prefer mildly cold skin over an irritating texture.

So it’s worth asking whether “aren’t you cold?” is the right question. Others may work better:

are you comfortable?

will you be running or sitting still?

would you rather wear the jumper or carry it?

These questions help children connect what they feel now with what they may need later.

When should parents insist?

Parents should be firmer when a child is very young, wet, unwell, exposed to strong wind or likely to remain outside for a long time.

Persistent shivering or numbness means it is time to get warm. Increasing clumsiness, unusual drowsiness, confusion or reduced responsiveness can indicate hypothermia, where the body’s core temperature has fallen dangerously low. Hypothermia is a medical emergency.

But for ordinary winter outings, flexible layers allow the plan to change. A jumper can be carried, added when activity slows, and removed when the child warms up again.

Your child reports the weather inside their body. You keep an eye on the weather outside it. A jumper in the bag lets your child feel heard, lets you keep the forecast in view, and lets everyone finally get out the door.

Joshua Pate, Associate Professor of Physiotherapy, University of Technology Sydney

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Few human experiences are as universal – or as poorly understood – as loneliness.

More than just a blue feeling and the inspiration for countless songs, it’s linked to poorer wellbeing, worse mental and physical health and a higher risk of early death.

Even in the age of social media, we hear of loneliness being on the rise everywhere – to the extent that some have labelled it an epidemic.

To build a better picture, we analysed responses to the Gallup World Poll from more than 218,000 people, across nearly 150 countries. We found around one in five people reported feeling lonely during a typical day.

But that global average masked a stark divide.

In high-income countries, 15.7% of people reported loneliness, including 16.2% in New Zealand and 17.8% in Australia. In low-income countries, by contrast, nearly one in three people (31.3%) reported feeling lonely.

The findings challenge the classic idea that loneliness is simply a private struggle. Instead, they suggest it is also a social condition, influenced by the economic circumstances, institutions and strength of community life.

The global picture

Across the data we analysed, one factor consistently stood out. That was the importance of social capital; the presence of trust and reliable support networks. People who report having others they can depend on are far less likely to feel lonely in every region of the world.

Other factors, such as poor health, unemployment and not having a partner, also played a part. People with more education tended to be less lonely in middle- and high-income countries. That relationship was much weaker in low-income countries, where wider social and economic conditions play a bigger role.

These findings led us to conclude that loneliness can often be the result of a person’s individual circumstances, together with their wider social and economic environment.

But was there evidence of a worsening global loneliness epidemic? While a popular idea, when we looked at data from 113 countries, between 2023 and 2024, we found no clear trend. In fact, just under two thirds of countries appeared to report lower rates in 2024 than 2023.

Again, this wasn’t uniform. While reported rates of loneliness rose significantly in low-income countries, they remained broadly unchanged in middle-income countries and declined in high-income countries.

Importantly, our findings should not be taken to mean loneliness is any less serious. It affects a substantial share of the global population and is strongly linked to poorer wellbeing, emotional distress and a range of adverse health outcomes.

Lessons for tackling loneliness

The consequences of loneliness extend far beyond feeling unhappy. Across all regions and income groups, people who reported loneliness tended to have lower life satisfaction; both in the present and in expectations for the future, though this association is smaller.

They were also far more likely to experience stress, worry, anger and physical pain, and much less likely to report enjoyment in their daily lives. While the strength of these relationships varied, the overall pattern was remarkably consistent across countries.

Our findings also suggest there is no single solution to loneliness. In low-income countries, where loneliness is both more common and increasing, reducing poverty, improving health systems and expanding social protection might all help people build stronger social connections.

In wealthier countries, where loneliness is less common but still affects millions of people, the challenge may be different. Investing in community organisations, public spaces and other opportunities for people to build trusted relationships could prove just as important as expanding mental health services.

More broadly, our findings suggest loneliness should be seen as more than an individual psychological experience. It also reflects the strength of the societies people live in.

That has implications that reach well beyond health. Tackling loneliness has the potential to improve wellbeing, boost productivity and help build stronger, more connected communities.

Dennis Wesselbaum, Associate Professor, Department of Economics, University of Otago and David Leblang, Professor of Politics and Public Policy, University of Virginia

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Australia is coming under increasing pressure from international drug companies demanding higher prices for their drugs.

In some cases, negotiations have broken down and companies have said they’ll withdraw their drugs from the market. In other cases, companies have said they’ll not apply for their drug to be listed on the Pharmaceutical Benefits Scheme (PBS) at all.

For example, the future of some PBS drugs for multiple sclerosis has been in doubt after pricing disagreements between the drug companies and the Australian government. These medicines will now remain on the PBS while a review of multiple sclerosis drugs takes place.

Earlier this year, drug company Eli Lilly said it would not apply to have its drug tirzepatide (Mounjaro) listed on the PBS for type 2 diabetes after price negotiations broke down.

So what’s really going on in the background? And where does that leave Australians’ access to subsidised medicines, now and into the future?

First the UK, then Australia?

United States drug companies have long sought higher prices for their drugs in international markets.

US President Donald Trump has stepped in to blame other countries for forcing drugs companies to accept lower prices.

We were subsidizing others’ healthcare […] it was really the countries that forced Big Pharma to do things that frankly, I’m not sure they really felt comfortable doing [that].

The US-UK trade deal on pharmaceuticals, announced in December 2025, is the first major surrender to this pressure.

In exchange for no tariffs on UK drug exports to the US, it requires the UK’s National Institute for Health and Care Excellence to increase the price it is willing to pay for new drugs. It also requires the National Health Service (NHS) to double its spending on new drugs by 2036.

The NHS is already squeezed for funds. Extra drug costs will mean cuts to other health care are estimated to lead to up to 291,000 extra deaths by 2036.

The National Institute for Health and Care Excellence currently approves more than 90% of new drugs it evaluates. So this deal will lead to higher prices, not many more new medicines.

It also creates a worrying precedent for other countries, including Australia.

But negotiations are not just about price

The first step to get a new medicine onto the Australian market is Therapeutic Goods Administration approval. Companies must provide clinical trial evidence of their drug’s effectiveness, either compared to a placebo or another drug. A new drug does not need to be better than existing treatments – just no worse.

Then, the company applies for their drug to be listed on the PBS, and the Pharmaceutical Benefits Advisory Committee – an independent team of experts that evaluates whether drugs are cost-effective and should be recommended for PBS listing – steps in.

It asks companies to compare their drug to standard treatment, in terms of health effects such as quality of life or preventing disability, hospitalisations or death. So price negotiations are not just about price. They are also about whether a new medicine improves health.

Sometimes there is no added health benefit. Analyses of the clinical trial evidence for more than 1,000 new medicine approvals in France found about half were no better and no worse than existing drugs.

If the Pharmaceutical Benefits Advisory Committee judges a new drug is no better, the allowable price is no higher than existing treatments for the same condition.

But not everyone’s happy.

Felicity McNeill, Chair of the advocacy group Better Access Australia, told the Australian:

In the past, big companies have been able to accept our pricing […] or even the massive price cuts agreed to by the industry, because we were just almost a charity case […].

Australia is hardly a “charity case”. We spend on average about A$1,250 per person per year on prescription and over-the-counter medicines, ranking eighth globally.

Australia subsidises a lower proportion of prescription medicines than many other countries in the Organisation for Economic Cooperation and Development (OECD) – 48% versus the OECD average of 59%.

But higher PBS prices for subsidised drugs will do little to increase this proportion. That’s because most unsubsidised drugs bought at the pharmacy are older, inexpensive medicines that cost less than a patient’s PBS co-payment. Even if these medicines were to be listed on the PBS, paying for them out-of-pocket would be cheaper.

Patients are weighing in

Patient groups have added their voices to the debate about the future of drugs they say are vital. They’ve shared their fears should multiple sclerosis drugs be removed from the PBS due to price negotiations breaking down.

However, there is little to no evidence of added benefits for these drugs, compared to the newer, more affordable treatment ublituximab (Briumvi). This has led to PBS recommendations for similar pricing.

Many patient groups depend on industry funding. And not all media reporting of patients’ perspectives mention all relevant details about price negotiations.

For example, in recent media reports on the multiple sclerosis drugs, patients do not mention manufacturers’ pricing as a barrier to access. Instead, the focus is only on the potential impact of the government’s price negotiations.

What should Australia do?

Pharmaceutical companies are highly profitable. And much research and development for new drugs is publicly subsidised. Public reimbursement of medicines is also key to equitable and affordable access. So the PBS should not give in to industry bullying.

International drug companies that do not apply for PBS listing or remove existing medicines from the market are likely to remain the exception, not the norm, as they will lose out on sales of their medicines.

If this does happen, there are other options available for Australia to access a needed drug. If a company is selling it elsewhere, Australia can buy it directly from another country via parallel imports.

Under exceptional situations, special access arrangements are also available for patients to access certain drugs that are not available in Australia.

We also need more transparency. Let’s open pricing decisions to public scrutiny. That way, Australians can know if multinational companies are crying poor because of unusually high development and production costs for a specific medicine, or whether these high prices just reflect what they think the market will bear.

Barbara Mintzes, Professor in Pharmaceutical Policy, School of Pharmacy and Charles Perkins Centre, University of Sydney; Ashleigh Hooimeyer, Associate Lecturer, School of Pharmacy, University of Sydney, and Kellia Chiu, Associate Lecturer, School of Pharmacy and Charles Perkins Centre, University of Sydney

This article is republished from The Conversation under a Creative Commons license. Read the original article.

The National Disability Insurance Scheme (NDIS) reform bill, now before parliament is the centrepiece of the federal government’s budget savings measures.

It’s the biggest reform the 13-year-old NDIS has faced and aims to ensure the scheme’s sustainability by bringing long-term funding growth down to 5–6% a year.

While a Senate committee inquiry completes its deliberations, the Greens and the cross-benchers are tinkering around the edges of the bill. They’re proposing minor amendments and small carve-outs from the harshest measures.

But they should be aiming higher.

The current bill includes short-term measures that are deeply concerning: a 50% cut to every NDIS participant’s social and community participation supports budget, and a 10% cut to every capacity-building daily activities budget.

These measures should be removed from the bill. These cuts will cause unnecessary harm to disabled Australians and, as our analysis shows, they’re unnecessary. The government can meet its objective of making the NDIS sustainable without them.

What’s been proposed?

An NDIS reform bill needs to pass to ensure the government can get on with necessary structural repairs to the scheme.

The bill includes:

• fixing flaws in the scheme’s design
• increasing consistency in assessing people’s eligibility and setting their budgets
• moving away from over-reliance on markets
• improving governance of the multi-billion-dollar program.

The bulk of the current bill relates to measures that will enable this repair.

What’s proposed for social and community participation budgets?

The government has also targeted social and community participation costs because they have grown faster than the scheme average. But they’re not the fastest-growing support category.

The “support determinations” section of the bill would be used to cut every NDIS participant’s social and community participation supports budget by 50%, and every capacity-building daily activities budget by 10%.

These cuts would be expected to stick: spending in these categories would be lower than it would otherwise be for years to come.

Social and community participation budgets support people to engage in everyday social, recreational, civic and community life and are central to the program’s purpose.

This might include travel training for an intellectually disabled person to build their independence in catching public transport to TAFE, sport or social activities. Or it might mean a support worker accompanying someone to attend a cultural celebration or community group.

What would cuts mean for participants?

These cuts will increase people’s social isolation and place additional burden on families and informal carers, mostly women, who would have to pick up the slack.

These cuts would also strip the most support from the most vulnerable people in the NDIS.

People in supported independent living or specialist disability accommodation, for example, have on average social and community participation budgets more than five times as high as other people in the NDIS. As such, they would lose five times as much support.

These cuts aren’t necessary to meet the government’s target

The government wants to bring NDIS cost growth down to 5–6% a year, from 11.3% in the 12 months to March 2026.

This is a reasonable ambition, and would be similar to growth in other social programs, such as Medicare and aged care.

But this target would be achieved with or without the proposed short-term cuts.

While the short-term cuts would deliver more than a third of the package’s A$37.8 billion savings over the four years to 2029–30, they would do nothing to dampen the long-term growth rate of the NDIS.

A one-off fall in costs

The government’s proposed short-term cuts would have an almost immediate effect on scheme costs, reaching their full impact of about A$4 billion a year by 2028–29.

But after 2028-29, savings from the short-term cuts would plateau. They would increase by only 5% in 2029-30, while savings from the structural reforms would increase by about 60%.

These growth patterns mean the short-term cuts would contribute a shrinking proportion of total savings from the reform package, down from 50% in 2026–27 to 26% in 2029–30.

This plateauing of savings means the impact of the short-term cuts would be a step fall in annual scheme costs, rather than a dampening of the rate of long-term growth.

Sustainability can be achieved without the short-term cuts

According to our calculations, over the next four years (the forward estimates), the full package of proposed reforms would drag NDIS cost growth down to 1.1% a year, a cut in real terms.

Without the short-term cuts, cost growth would be 3% a year. That would still be a real cut in population-adjusted terms, and is well below the government’s target long-term NDIS growth rate of 5–6% a year.

The government has not made the case for the necessity for such deep and blunt cuts in the short term.

In the period after 2029–30, NDIS cost growth under the full package of reforms would be about 5%, from a base cost of $56.2 billion in 2029–30.

Were the bill to be passed with the short-term cuts removed, cost growth would still be about 5%, but from a $4.3 billion higher base cost of $60.5 billion in 2029–30.

Bottom line

Cuts to social and community participation budgets achieve no policy goal, and do nothing to make the NDIS sustainable, while hurting people who rely on these supports to participate in their communities.

The government’s insistence on making these cuts is about shoring up the budget bottom line – but at the expense of some of the most vulnerable Australians. A similarly blunt approach to Medicare or hospital funding would be unthinkable.

Passing the bill with these cuts included would hit the short-term savings target but entirely miss the point.

Sam Bennett, Disability Program Director, Grattan Institute and Owain Emslie, Senior Associate, Disability Program, Grattan Institute

This article is republished from The Conversation under a Creative Commons license. Read the original article.

In the United States, Google’s (now Alphabet) Debug initiative has asked the federal government for permission to release up to 32 million sterilised male mosquitoes in California and Florida.

Male mosquitoes don’t bite or carry disease. The hope is these millions of male mosquitoes will mate with disease-carrying females, the eggs won’t hatch, and mosquito numbers will fall.

Unsurpringly, the US public has questions about the plan, including whether the release is safe and will work at scale, and what happens afterwards. Google/Alphabet’s plan is to release more than 16 million mozzies each year, for two years.

More than a decade ago, I led a project with Google’s life sciences division, (now known as Verily), to test a novel mosquito-control strategy in far north Queensland.

The results showed releasing specially bred male mosquitoes can dramatically suppress populations of the exotic, invasive mosquito species Aedes aegypti. This species is responsible for spreading deadly diseases such as dengue, Zika, chikungunya and yellow fever. Given the US is seriously considering this approach, our Australian experience offers important lessons.

Using mosquitoes against themselves

This story begins in 2015, when we travelled to Silicon Valley to meet Verily scientists interested in developing a mosquito suppression technology. Their goals aligned closely with our own research, supported by an Australian National Health and Medical Research Council grant, to develop environmentally friendly tools for suppressing invasive mosquitoes.

The strategy focuses on male mosquitoes because they don’t bite, and female Aedes aegypti, which generally only mate once during their lifetime. If that mating event was incompatible – meaning the embryos don’t develop – it could not produce viable offspring. Our challenge was to make mating ineffective.

The approach we ultimately tested relied on Wolbachia, a naturally occurring bacterium found in many insects. Some Wolbachia strains create a form of reproductive incompatibility, as described above.

The theory is simple: release enough Wolbachia-carrying males into a population and, over time, the population declines. The released males are also beautifully evolved to search and find the last females – their large bushy antennae are super-radars for this job.

North Queensland the perfect laboratory

The Cassowary Coast in north Queensland provided ideal conditions for a large-scale trial of this approach. The region contained towns with abundant Aedes aegypti populations, while surrounding agricultural areas limited movement between communities. Equally important was the support of residents, local government and First Nations leaders.

Aedes aegypti likely arrived in Queensland in the late 1800s. It is distinct from our native mosquitoes because it is highly domesticated and feeds mainly on humans.

Before a single mosquito was released, the project team spent two years conducting field surveys and engaging with communities. We met with households, community organisations, First Nations leaders and local councils to discuss the technology and answer questions. The project, known as “Debug Innisfail”, ultimately received regulatory approval from multiple authorities.

Releasing three million male mosquitoes

Field surveys began in 2015, involving a collaborative team spanning Australian and US institutions. During a 20-week release period in 2018, around three million Wolbachia-carrying male mosquitoes were released into three treatment towns. Meanwhile, control towns where no mosquitoes were released were monitored.

The release system itself reflected Verily’s engineering strengths. The company developed bespoke technologies, including machine-learning-based systems capable of separating male and female mosquitoes, crucial to ensuring only males were released.

The results were striking: when compared with control towns, mosquito populations in towns where mosquitoes were released began declining within four weeks. The findings, published in 2021, demonstrated incompatible male mosquito releases could achieve strong suppression.

In two treatment towns, suppression effects persisted into the following year. In one town, monitoring detected only a handful of Aedes aegypti 12 months later, corresponding to roughly 95% suppression.

What this means for the US

The Australian trials provide evidence-based answers to many of the concerns now being raised in the US.

First, ecological impacts are likely to be very small. Aedes aegypti is an invasive species in Australia and many other countries. Because it exclusively lives around humans and bites them, removing it from urban environments has minimal ecological consequences.

Second, the approach can work at scale. Although adult mosquitoes survive for only a few days, continuous releases of highly competitive males can substantially reduce populations across entire towns.

Third, benefits may persist after releases finish. This is because the suppression outcome does not necessarily disappear straight away, and can carry over into subsequent seasons. But that doesn’t mean mosquito biology can be ignored – success depends on factors such as local ecology, mosquito movement patterns, and community participation. The technology alone is not enough.

A model for future mosquito control

Perhaps the most important lesson from the trials is the value of collaboration. This project brought together researchers from six universities and Verily. Combining scientific expertise with industrial-scale engineering accelerated the journey from laboratory concept to real-world field trial in an incredibly short time.

We are still working towards biological and mechanical approaches to efficiently separating male mosquitoes, which would have better utility in developing countries.

As Aedes aegypti expands its range and insecticides fail to suppress it, using the mosquito against itself as the biological control tool will become increasingly important.

The Queensland trials helped lay the groundwork for programs now underway in the US. And they are a reminder that when science, technology and communities work together, it is possible to solve problems that matter.

Nigel Beebe, Professor, The University of Queensland

This article is republished from The Conversation under a Creative Commons license. Read the original article.

As we age, low back pain becomes more common. Between the ages 20 and 59, persistent low back pain (lasting more than three months) affects nearly one in five.

While you may be tempted to reach for heat packs, medication or a massage, new evidence suggests that the common advice to go for a swim is much more effective.

Until recently, there was no research to back up this advice. Our new trial shows for the first time that swimming can improve low back pain.

What we did

We recruited 76 adults aged 26 to 74 who experienced persistent and bothersome low back pain for more than three months. They needed to be able to swim 25 meters independently, but didn’t swim regularly.

The participants were randomly allocated to receive either a swimming and education program (the intervention group) or education only (the control group).

The intervention involved an eight-week individualised swimming program, supported by four telehealth sessions with a physiotherapist, and free access to a local indoor or outdoor swimming pool.

The amount of swimming was tailored to each participant’s ability and fitness level, with the goal of completing three 30–45 minute swimming sessions per week by the end of the program. Participants were encouraged to continue swimming after the program completed.

The education, provided by a physiotherapist, aimed to help them better understand their pain, reduce the fear associated with movement and exercise, and increase confidence to manage their back pain.

While many people with back pain believe they should avoid activity to protect their back, a large body of evidence shows remaining active is better.

Participants in the education-only (control) group had one to two sessions with a physiotherapist to cover the same key messages about back pain, but otherwise continued their usual treatment and activity.

What did we find?

Participants in the swimming group reported improved function, less pain and more confidence to manage future back pain.

Their disability reduced by more than 50% at the end of the eight-week program. This could mean that a person improved their ability to do daily activities such as standing from a chair, walking or sleeping.

The swimming group’s improvements were 30% greater than those who received education only (control) group.

Participants told us swimming appealed to them because it was low-impact, it reduced weight-bearing and strain, and it enabled them to exercise more confidently and with less pain. They also reported additional health benefits, including better mood.

Swimming can be used as a way for people with back pain to start exercising and break the cycle of pain and limited activity.

What happened after the program ended?

At the end of the eight-week program, some people kept swimming, others swapped to another form of exercise and some stopped exercising.

Although most participants reported enjoyed swimming, some found accessing a pool and the time required a barrier to keep going in the longer term.

Twelve months after starting the program, there were still benefits for the swimming group for disability, function and confidence. But the difference between groups became smaller over time.

What were the limitations?

Our study’s sample size was relatively small, so these findings need to be confirmed in future, larger studies.

Further trials are also needed to test whether the results still hold for people with more severe or disabling back pain.

Finally, the volunteers in our study knew we were investigating swimming and had positive expectations of swimming before starting. This is a limitation that could affect the findings.

How does swimming compare to other exercises for back pain?

A range of different exercises – including Pilates, functional exercises (a type of strength training that helps you perform daily activities) and structured walking – have been shown to be beneficial for treating disability and preventing low back pain recurrences.

While we didn’t compare swimming to another type of exercise, the benefits we identified were as large or larger than previously reported for other exercises.

People with chronic low back pain can now consider swimming as an evidence-based exercise option and be more confident in choosing it as part of their long-term management.

But if you find it a hassle to get to a pool, or don’t like swimming, it may help get your back pain under control before moving to a type of exercise you prefer or can more easily access.

Mark Hancock, Professor of Physiotherapy, Macquarie University and Deborah Wareham, Postdoctoral Research Fellow, Spinal Pain Research Centre, Macquarie University

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Across Australia, “longevity” and “wellness” clinics, most of them online, sell injections that promise to slow ageing, build muscle, heal injuries and help you sleep.

This week’s ABC 4 Corners described a doctor prescribing a “stack” of them after a short questionnaire and a phone call. Many are peptides.

A peptide is a short chain of amino acids: the tiny building blocks of the proteins in our bodies. A few peptides, such as insulin, are approved medicines.

But the peptides sold at online wellness clinics – with names such as BPC-157, CJC-1295 and TB-500 – aren’t approved, nor are most other “anti-ageing” injections.

None has been shown to have the claimed effects in humans, and the medicines regulator warns they carry real risks.

Some people import a supply for their own use. For some peptides, that is lawful with a valid prescription.

Others are controlled imports that also need an importation permit a patient cannot get. Without the right authority, or if the peptides are sold on, importing them is an offence.

And it doesn’t matter if the substances are labelled “research chemicals, not for human use” if, in substance, they are therapeutic goods. Australian law does not recognise any “research chemicals” exemption.

But others get these substances via online clinics that exploit a legal loophole: these injections are “unapproved” but that doesn’t mean they’re “banned”.

What the medicines regulator does

Before a medicine can be sold in Australia, it usually has to be listed on an official register. To get there, the Therapeutic Goods Administration (TGA) – the government body that regulates medicines – must check it is safe, well made and works.

The TGA regulates according to risk, putting its limited resources where the danger is greatest.

But even a real, known risk may draw no regulatory action because it’s another regulator’s job or because acting is resource-intensive, costly and may not achieve the aim of making patients safer.

Either way, the peptides mentioned above have never passed that check. In other words, they’re “unapproved”.

How the clinics get around the rules

Doctors have always been allowed to supply unapproved products in special cases, through the Special Access Scheme and other lawful access pathways.

That way, a patient with a rare cancer, for example, or a child who cannot take an approved medicine, still has options.

But some Australian clinics exploit a different exception. When a pharmacist mixes up a medicine intended for one patient – called compounding – that preparation doesn’t need to be on the register.

The rule was written for cases like a pharmacist making a liquid version of a tablet for someone who cannot swallow.

Making the same recipe in bulk and posting it interstate as injection pens is really manufacturing, which needs a licence.

But because each dose is labelled “for a particular person”, it never faces the TGA’s safety check.

Two regulators, one gap

Australia splits the job of regulating supply of medicines across two regulators, plus consumer law, which can act against misleading claims.

The TGA oversees therapeutic products. It:

• decides what goes on the register
• licenses the manufacturers
• polices the advertising
• can fine or prosecute a business.

But it doesn’t regulate doctors or pharmacists. That falls to the Australian Health Practitioner Regulation Agency (Ahpra), which is responsible for a practitioner’s registration.

Ahpra can attach conditions to a practitioner’s registration, suspend them, or take steps to de-register them. But this usually only happens after someone complains, one practitioner at a time.

Ahpra has recently tightened its rules on cosmetic and online prescribing by doctors, so a script can’t be written after just completing a questionnaire. It has also tightened rules for compounding by pharmacists.

But online clinics can slip through a regulatory gap. If the compounded product is notionally exempt, the TGA may not intervene, while the acts of prescribing and compounding fall to Ahpra, which works case by case and can therefore be slow.

Can these injections hurt you?

Yes. The TGA warns you often cannot be sure what is in the vial, whether the dose is right, or whether it is sterile.

Even when pure, these drugs can harm. A single dose of CJC-1295 raises a growth factor, IGF-1, up to three-fold for over a week. High IGF-1 levels are a probable cause of breast cancer.

Peptides and anti-ageing injections are also very costly. A few weeks’ supply can cost more than A$1,000.

When it goes wrong, who’s accountable?

The clearest wins against these businesses have been over their advertising, rather than harm to a patient.

In 2019, the Federal Court fined one company, Peptide Clinics, A$10 million for advertising prescription peptides to the public. It found no real doctor–patient relationship behind the online sales. With a real clinical relationship, the sales would have counted as clinical advice, which the advertising ban does not prohibit.

The company collapsed during the case.

These clinics depend on advertising, and advertising prescription medicines to the public is against the law. So an advertising case can be brought against the whole business, not just one doctor.

Penalties run up to $1.65 million a breach for a person or $16.5 million for a company. But the TGA rarely goes that far.

It usually starts small, warning and educating first, then escalating if that fails.

The TGA also says it cannot chase every breach.

This week, the TGA put suppliers and promoters on notice, admitting repeated warnings have not stopped the peptide trade.

So what needs to change?

We probably don’t need new laws. Refinement of existing laws would do. The options include:

• enforcing the ban on the advertising these businesses depend on

• funding the TGA properly so it can pursue breaches

• narrowing the compounding rule so it cannot work as a bulk manufacturing licence.

The government did this in 2024, removing weight-loss drugs like Ozempic (by then, sold to at least 20,000 Australians by online clinics compounding copies at scale), along with gene therapies and medicinal cannabis. It could do the same for these peptides.

The patients buying these injections are not reckless. They are trying to look after their health. But they may be relying on a promise the clinical evidence cannot support.

Christopher Rudge, Lecturer in Law, Sydney Law School, University of Sydney

This article is republished from The Conversation under a Creative Commons license. Read the original article.