In the flurry of action in Parliament House in the final moments of the sitting year, the government passed a bill that escaped the attention of most.

New changes to social security law mean a person’s income support can now be cancelled because they are subject to an outstanding arrest warrant for a serious offence.

These are people merely accused of crimes, not found guilty of them.

The change raises fundamental questions about justice, human rights and the role of social security. It transforms welfare from a crucial safety net to a tool of law enforcement, with serious implications.

Flying under the radar

The changes were quietly added to Social Security and Other Legislation Amendment Bill in late October. The government did this without consultation or announcement, and bypassing parliamentary committee scrutiny on the grounds of urgency.

Less than a month later, the amended bill was passed into law. The change was effective immediately.

The legislation grants the minister for home affairs the power to authorise “Benefit Restriction Notices”. These cancel social security, family assistance and parental leave payments for anyone with an outstanding arrest warrant for serious, violent or sexual offences (within the meaning of the criminal code).

No conviction or even court appearance is required.

According to the bill’s explanatory memorandum:

the objective […] is to ensure people who are subject to an outstanding arrest warrant for a serious offence can no longer be supported through the social security and family payments systems.

But an arrest warrant is not proof of guilt. It is merely an allegation that someone may have committed an offence.

Yet curbing social security is a punishment, and not just for the suspect, but often also their family.

6 serious problems

1. Violation of the presumption of innocence

These laws would enable punishment before any judicial determination of guilt. A person could have their support payments cancelled even if they haven’t been charged, convicted or appeared before any court.

The parliament’s own human rights experts warn this may punish people who are legally innocent, directly contradicting the presumption of innocence.

This presumption is expressly protected under the International Covenant on Civil and Political Rights, to which Australia is a party. Its erosion under the measure sets a dangerous precedent.

It also cuts across the fundamental protection against double jeopardy: the principle that people will not be tried or punished twice for the same offence. This is because cancelling welfare is a form of punishment, after which a second punishment might also be enforced for those subsequently found guilty.

2. Ministerial power with few safeguards

The laws bypass normal checks and balances in Australia’s social security system. The minister can cancel payments based on police requests, with no independent review.

Ordinarily, decisions by Services Australia can be appealed to the Administrative Review Tribunal. But under this measure, only limited judicial review is available. Courts can check procedural issues, but not whether the decision was fair.

Payments may be cancelled without the person knowing a warrant exists and there is no obligation to reinstate benefits if the warrant is cleared or charges dropped. Back pay isn’t provided if the person is found innocent.

This concentration of power removes safeguards against error and abuse, creating a two-tier system that denies basic procedural protections.

Constitutionally, it blurs the separation of powers designed to ensure courts, not politicians, decide guilt and punishment.

3. First Nations peoples will be hardest hit

First Nations peoples make up 3.8% of the population, but 36% of all prisoners, with this overrepresentation continuing to grow. This measure will hit First Nations communities hardest.

The experience of similar powers in Aotearoa/New Zealand since 2013 has shown Māori peoples have their social security payments cancelled at twice the rate of others.

As a result, the nation’s Welfare Expert Advisory Group recommended removing the powers in 2019.

By 2019-2020, 71% of warrant to arrest sanction recipients were Māori.

And while the Australian police say its use of this power will be rare, the similar laws in NZ were used around 700 times in 2019, according to the latest available data.

4. Harmful impacts for domestic violence victims

Domestic violence victims are also at significant risk from the measure. Victims fleeing abuse – especially First Nations victims – are increasingly wrongly identified as perpetrators.

If an arrest warrant is issued while they are in hiding, their support payments could be cancelled, cutting off their income at the most dangerous moment in their lives with no chance to explain or present evidence.

This potentially forces them back into violent situations, with the alternative “choice” being dire poverty.

5. Serious doubt about proportionality and effectiveness

These laws could only comply with international law if the measure is proportionate, and actually effective in its objective of stopping payments to people with outstanding arrest warrants “which might be assisting them in evading the authorities”.

The government has provided no evidence that cutting off payments would prevent people from evading the law or encourage their surrender.

That the measure cancels, rather than suspends, payments is also arguably in contradiction with international law, given this less restrictive alternative is available.

6. Another legal problem in the making?

Australia should have learned from its Robodebt Royal Commission. Welfare cancellations without proper safeguards can be found unlawful and cause devastating harm.

Yet, the Benefit Restrictions Notice regime risks creating conditions for another scandal.

When the United States introduced similar “fugitive felon” provisions in 1996, they proved disastrous, with many elderly and vulnerable people losing benefits without knowing warrants existed.

By 2002, around 110,000 people had their benefits removed under these provisions.

Following legal challenges and a class action settlement, the US severely restricted these measures and compensated millions of dollars to people whose benefits were wrongly cancelled.

What needs to happen

While these laws are now active, their real-world consequences will take time to unfold.

It remains to be seen whether they will facilitate arrests. In the meantime, there must be rigorous public reporting, independent scrutiny, and formal review of how these powers are used, to ensure the serious risks outlined here do not materialise unchecked.

Zoe Staines, Senior Lecturer in Law and Social Policy, The University of Queensland; Francis Markham, ARC DECRA Fellow, Australian National University; Hannah McGlade, Associate Professor in Law, Curtin University, and Thalia Anthony, Professor of Law, University of Technology Sydney

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

The question of when people first arrived in the land mass that now comprises much of Australasia has long been a source of scientific debate.

Many Aboriginal people believe they have lived on the land since time immemorial. But until the advent of radiocarbon dating techniques, many western scholars thought they had arrived not long before European contact 250 years ago.

Now a new study by an international collaboration of geneticists and archaeologists, including myself, suggests that humans first arrived in Sahul – the “super-continent” that encompassed New Guinea and Australia during the last ice age – by two different routes around 60,000 years ago.

The research, led by archaeologist Helen Farr at the University of Southampton, also points to the earliest uncontested example of travel by boat – probably simple watercraft such as paddled bamboo rafts or canoes. The first people to arrive would have migrated into the region following a rapid dispersal from Africa around 10,000 years earlier.

The key to the work of our genetics team, based at the University of Huddersfield, is mitochondrial DNA (mtDNA). People only inherit mtDNA from their mothers, so we were able to track an unbroken maternal line of descent down many generations, during which the mtDNA gradually accumulates small mutations.

We sequenced mtDNA genomes in almost 1,000 samples, mainly from New Guineans and Aboriginal people – collected by colleagues at La Trobe University in Melbourne and the University of Oxford, in close collaboration with the communities.

The samples were all collected with the help of Aboriginal elders. The principal elder, Lesley Williams from Brisbane, arranged invitations for the researchers to address Aboriginal groups to explain the purpose of the study and answer any questions before signed consent was given. The results of the analysis of each sample were returned in person whenever possible.

These genealogical trees were then combined with another 1,500 sequences that were already available. By counting the number of mutations from ancestors in these trees, we could use a “molecular clock” to date lineages that were unique to New Guineans, Aboriginal people or both.

After correcting for natural selection (which makes the mutation rate non-linear) and checking the results against well-known colonisation events in the Pacific, we concluded that the deepest lineages were 60,000 years old. Reanalysing previously published male-lineage and genome-wide data found that this also fitted with our results.

Clashing chronologies

The debate about when and how people first arrived in modern-day Australasia was transformed during the 20th century, especially by the introduction and gradual refinement of radiocarbon dating techniques.

This pushed the time of people’s first arrival back to around 45,000 years – ironically, now known as the “short chronology”. However, some archaeologists argued they may have arrived even earlier.

In 2017, newer scientific dating methods – such as optical luminescence dating, which estimates the time quartz grains in the sediments embedding human remains were last exposed to sunlight – supported the so-called “long chronology” of people first arriving in northern Australia at least 60,000 years ago. But this view remained contentious.

The pendulum swung again in 2024, as geneticists weighed in with a genetic clock based on the recombination that takes place between pairs of chromosomes with every generation. New results using this clock suggested that interbreeding between early modern humans and Neanderthals, shortly after modern humans left their African homeland, took place less than 50,000 years ago – more recently than had previously been proposed.

All present-day non-Africans carry around 2% Neanderthal DNA, suggesting they must all be descended from that small group. This research therefore supported the short chronology view.

The genetic and archaeological evidence could apparently only be squared if there had been a first wave of early arrivals in Sahul at least 60,000 years ago, that was entirely replaced by a second wave of modern humans around 40,000 years ago. For some experts this seemed implausible, since people were already widespread in Sahul by that time.

Our genetic dates suggest a simpler solution. There was only one wave 60,000 years ago, and these earliest arrivals were the ancestors of today’s New Guineans and Aboriginal people in Australia.

The earliest seafarers

Our results suggest there were two distinct migrations into Sahul – both around the same time about 60,000 years ago. This is because the most ancient lineages fell into two groups.

The major set, with ancestry in the Philippines, was distributed throughout New Guineans and Aboriginal people in Australia. But we also identified another minor set, with ancestry in South Asia or Indochina, only in Aboriginal people. The simplest explanation for these patterns is that there were two dispersals into Sahul: a major northern pathway and a minor southern route.

Both groups of migrating people met more archaic species of human along the way. As well as the 2% Neanderthal DNA that all non-Africans carry, the genomes of modern New Guineans and Aboriginal people in Australia carry a further 5% of archaic human DNA with more local origins – the results of interbreeding in Southeast Asia and perhaps even in Sahul itself.

Even with the lower sea levels 60,000 years ago, that second group must have crossed at least 60 miles (100km) of open sea to reach Sahul – some of the earliest evidence we have for human seafaring. An increasing amount of research suggests maritime technology played a role in early humans’ rapid dispersal from Africa some 10,000 years earlier, taking a coastal route via Arabia to Southeast Asia and beyond.

But the debate about precise timings of these earliest journeys doesn’t end here. We are now analysing whole human genome sequences – each consisting of 3 billion base units, compared with 16,500 for mtDNA – to further test our results. But both kinds of genetic clock – the mutation clock we use, and the recombination clock advocated by others – are indirect evidence. If ancient DNA can eventually be recovered from key remains, we can test these models more directly.

It may happen. Recovering ancient DNA from the tropics is challenging, but in the rapidly evolving world of archaeogenetics, almost anything now seems possible.

Martin B. Richards, Research Professor in Archaeogenetics, Department of Physical and Life Sciences, University of Huddersfield

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

Defence Minister Richard Marles has confirmed Australia is monitoring a flotilla of Chinese Navy ships currently in the Philippine Sea but with its destination unknown.

Marles volunteered the information while announcing a shakeup that will establish a new Defence Delivery Agency designed to improve military acquisition and sustainment operations.

The agency will be headed by a national armaments director, who will advise the government on strategies for acquisitions and the delivery of projects after they have been approved. The government says it is the biggest reform in defence organisation in half a century.

Marles, who is acting prime minister while Anthony Albanese is on his honeymoon this week, went out of his way to say the Chinese ships were being tracked, after a report about them in the Australian Financial Review last week.

He told a news conference the government did not yet have a sense of where the task group was going. “But we continue to monitor it as we monitor all movements until we know that the task groups are not coming to Australia.”

Earlier this year, the government was caught out when Chinese ships conducted a live-fire exercise in the Tasman Sea. A Virgin pilot sounded the alarm.

According to some sources, the Defence Department had alerted the government to that flotilla, but the government had decided not to say anything publicly, only to be thrown onto the back foot when the issue blew up. The flotilla later sailed around Australia.

Marles said on Monday:

We’re not about to give a running commentary on the movements of all Chinese Navy vessels, but in light of the report that was made on Thursday, we thought that it was important to make these statements and to make them in the proper context. So that Australians can be assured that we are monitoring our areas of interest and we are monitoring the movements of the Chinese Navy.

The change to acquisition advice and oversight is a reflection of discontent over a long period with the Defence Department. Defence projects have been notoriously behind time and over budget.

Marles said the new agency would be independent. It will report directly to the ministers of defence and defence industry.

It will begin operations on July 1 when three existing groups will be merged – the Capability Acquisition and Sustainment Group, the Guided Weapons and Explosive Ordinance Group, and the Naval Shipbuilding and Sustainment Group. The new independent entity will then become the Defence Delivery Agency on July 1 2027.

Marles said the establishment of the new agency “will see a much bigger bang for buck for the defence spend. And that is at the heart of the decision that we have made. It puts a focus on delivery and will ensure that it is much more sharp in the way in which it is undertaken.

"It will mean advice comes to government much earlier in the process about the challenges that are facing any particular program, any particular project, so that we can ensure those projects are delivered on time and on budget.”

Opposition defence spokesman Angus Taylor said the announcement was a matter of moving bureaucrats around. There was no increase in funds, he said.

Michelle Grattan, Professorial Fellow, University of Canberra

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

Today, the Albanese Labor government released the long-awaited National AI Plan, “a whole-of-government framework that ensures technology works for people, not the other way around”.

With this plan, the government promises an inclusive artificial intelligence (AI) economy that protects workers, fills service gaps, and supports local AI development.

In a major reversal, it also confirms Australia won’t implement mandatory guardrails for high-risk AI. Instead, it argues our existing legal regime is sufficient, and any minor changes for specific AI harms or risk can be managed with help from a new A$30 million AI Safety Institute within the Department of Industry.

Avoiding big changes to Australia’s legal system makes sense in light of the plan’s primary goal – making Australia an attractive location for international data centre investment.

The initial caution is gone

After the public release of ChatGPT in November 2022 ushered in a generative AI boom, initial responses focused on existential risks posed by AI.

Leading AI figures even called for a pause on all AI research. Governments outlined plans to regulate.

But as investment in AI has grown, governments around the world have now shifted from caution to an AI race: embracing the opportunities while managing risks.

In 2023, the European Union created the world’s leading AI plan promoting the uptake of human-centric and trustworthy artificial intelligence. The United States launched its own, more bullish action plan in July 2025.

The new Australian plan prioritises creating a local AI software industry, spreading the benefit of AI “productivity gains” to workers and public service users, capturing some of the relentless global investment in AI data centres, and promoting Australia’s regional leadership by becoming an infrastructure and computing hub in the Indo-Pacific.

Those goals are outlined in the plan’s three pillars: capturing the opportunities, spreading the benefits, and keeping us safe.

What opportunities are we capturing?

The jury is still out on whether AI will actually boost productivity for all organisations and businesses that adopt it.

Regardless, global investment in AI infrastructure has been immense, with some predictions on global data centre investments reaching A$8 trillion by 2030 (so long as the bubble doesn’t burst before then).

Through the new AI plan, Australia wants to get in on the boom and become a location for US and global tech industry capital investment.

In the AI plan, the selling point for increased Australian data centre investment is the boost this would provide for our renewable energy transition. States are already competing for that investment. New South Wales has streamlined data centre approval processes, and Victoria is creating incentives to “ruthlessly” chase data centre investment in greenfield sites.

Under the new federal environmental law reforms passed last week, new data centre approvals may be fast-tracked if they are co-located with new renewable power, meaning less time to consider biodiversity and other environmental impacts.

But data centres are also controversial. Concerns about the energy and water demands of large data centres in Australia are already growing.

The water use impacts of data centres are significant – and the plan is remarkably silent on this apart from promising “efficient liquid cooling”. So far, experience from Germany and the US shows data centres stretching energy grids beyond their limit.

It’s true data centre companies are likely to invest in renewable energy, but at the same time growth in data centre demands is currently justifying the continuation of fossil fuel use.

There’s some requirement for Australian agencies to consider the environmental sustainability of data centres hosting government services. But a robust plan for environmental assessment and reporting across public and private sectors is lacking.

Who will really benefit from AI?

The plan promises the economic and efficiency benefits of AI will be for everyone – workers, small and medium businesses, and those receiving government services.

Recent scandals suggest Australian businesses are keen to use AI to reduce labour costs without necessarily maintaining service quality. This has created anxiety around the impact of AI on labour markets and work conditions.

Australia’s AI plan tackles this through promoting worker development, training and re-skilling, rather than protecting existing conditions.

The Australian union movement will need to be active to make the “AI-ready workers” narrative a reality, and to protect workers from AI being used to reduce labour costs, increase surveillance, and speed up work.

The plan also mentions improving public service efficiency. Whether or not those efficiency gains are possible is hard to say. However, the plan does recognise we’ll need comprehensive investment to unlock the value of private data holdings and public public data holdings useful for AI.

Will we be safe enough?

With the release of the plan, the government has officially abandoned last year’s proposals for mandatory guardrails for high-risk AI systems. It claims Australia’s existing legal frameworks are already strong, and can be updated “case by case”.

As we’ve pointed out previously, this is out of step with public opinion. More than 75% of Australians want AI regulation.

It’s also out of step with other countries. The European Union already prohibits the most risky AI systems, and has updated product safety and platform regulations. It’s also currently refining a framework for regulating high-risk AI systems. Canadian federal government systems are regulated by a tiered risk management system. South Korea, Japan, Brazil and China all have rules that govern AI-specific risks.

Australia’s claim to have a strong, adequate and stable legal framework would be much more credible if the document included a plan for, or clarity about our significant law reform backlog. This backlog includes privacy rights, consumer protection, automated decision-making in government post-Robodebt, as well as copyright and digital duty of care.

Ultimately the National AI Plan says some good things about sustainability, sharing the benefits, and keeping Australians safe even as the government makes a pitch for data centre investment and becoming an AI hub for the region.

Compared with those of some other nations, the plan is short on specificity. The test will lie in whether the government gives substance to its goals and promises, instead of just chasing the short-term AI investment dollar.

Jake Goldenfein, Senior Lecturer, Law and Technology, The University of Melbourne; Christine Parker, Professor of Law, The University of Melbourne, and Kimberlee Weatherall, Professor of Law, University of Sydney

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

Thrush is one of the most common infections in the world. It’s caused by the fungi Candida – specifically, the yeast Candida albicans. Although yeast infections are normally treated easily with antifungal drugs, a growing number of Candida species are developing resistance to these drugs – including the species that causes thrush.

According to the US Centers for Disease Control and Prevention, about 7% of all Candida blood samples tested are resistant to the antifungal drug fluconazole, the first-line drug used to treat most Candida infections.

This means there are fewer treatment options for even routine thrush infections – making them more difficult to treat. It also means that more severe Candida infections, which can occur in people who have a weakened immune system or are taking long courses of antibiotics, will become even harder to manage.

Antifungal resistance may also be contributing to the rise in recurrent thrush (thrush infections which continue to come back). This affects around 138 million women worldwide, but is expected to rise to 158 million people by 2030.

Why resistance is growing

The antifungal resistance landscape has changed dramatically over the past few decades.

In the early–to-mid 2000s, antifungal resistance was rare. Fluconazole worked well for most Candida albicans infections, less than 5% of which were resistant to it.

But Candida albicans is a highly adaptable microorganism, which can easily develop resistance to antifungals under the right conditions.

Research shows that resistance among Candida albicans has been trending upwards over the past eight years at least. A small study of patients in Egypt found that in 2024, nearly 26% of Candida albicans isolates from blood samples were resistant to fluconazole. However, more research is needed to understand whether this picture is the same worldwide.

Candida can develop resistance to antifungal drugs through genetic mutations which make them less susceptible to antifungals, or help them reduce the drug’s effectiveness.

Candida can also protect itself from antifungal drugs by forming tough biofilms. These slimy layers of fungal cells block drugs from getting in, help the fungus pump any drugs which have penetrated the barrier back out, and allow some cells to hide in a resting state until treatment is over. Candida can also alter the structure of molecules targeted by antifungals in order to prevent the drugs from binding effectively.

The key reason Candida infections are becoming harder to treat is because the fungi are adapting to survive antifungal drugs. But this resistance isn’t happening by chance. There are several factors that are contributing to the problem, including misuse and overuse of antifungal drugs (not just by people but in agriculture too) and the limited number of effective antifungal drugs that are available (which are difficult and expensive to develop).

Increasing environmental temperatures, ecological stress and fungicide use are also creating conditions that favour heat-tolerant and drug-resistant Candida strains – such as Candida auris, which is highly resistant to multiple classes of antifungal drugs, and can cause severe infection in people who have a weakened immune system.

Preventing antifungal resistance

Candida is primarily transmitted through person-to-person contact, sexual contact and contact with contaminated objects or surfaces. In healthcare settings, Candida can also spread through contaminated medical equipment and devices.

Airborne transmission is not common with Candida. However, an alarming recent study reported that species of Candida resistant to common antifungal drugs were detected in urban air samples in Hong Kong. This included Candida albicans.

The presence of Candida in air could increase the likelihood of community spread and elevate the risk of inhalation – particularly in hospitals, crowded areas or care homes with immunocompromised people. This represents a potential route of exposure that has previously been underestimated. More studies will be needed to investigate where urban Candida originates and how infectious it may be.

Candida generally doesn’t cause harm under normal conditions and if you have a healthy immune system. Maintaining a healthy micriobiome is key to protecting yourself: the beneficial bacteria in your body help keep Candida levels under control and prevent it from overgrowing and becoming problematic.

However, when the balance of your friendly bacteria is disrupted – for example, by antibiotics, poor diet, a weakened immune system or high stress – Candida can grow out of control, leading to illnesses.

Microbiome disruption can also create conditions where antifungal-resistant Candida can overgrow, form resistant biofilms and become harder to treat.

Looking after your microbiome can make a significant difference in reducing the risk of Candida and other infections. This involves eating a diverse, fibre-rich diet – including fermented foods – and limiting highly processed foods.

Only take antibiotics when prescribed. Probiotics and prebiotics may also help maintain your microbiome balance, especially after antibiotic use or recurrent infections.

While most Candida infections are treatable, drug-resistant strains and infections in vulnerable people can be serious. However, we can all do our part to prevent resistant strains from developing – including by only taking antifungal medicines exactly as prescribed, completing the full course, and maintaining good hygiene.

Manal Mohammed, Senior Lecturer, Medical Microbiology, University of Westminster

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

This year, a series of rallies organised by neo-Nazi groups in Australian cities sparked public outrage and concern about the extreme right.

Yet, some media coverage of the rallies downplayed the role neo-Nazis played in what they called “anti-immigration rallies”. Other commentators misrepresented statistics on net migration.

Politicians, meanwhile, traded barbs about who was to blame for far-right demonstrators on city streets.

In the United States, there was a similarly muddled response to a recent scandal involving genocidal, racist text messages among young Republican leaders.

The messages included racist slurs, praise for Adolf Hitler and jokes about gas chambers. Yet, Vice President JD Vance dismissed them as “edgy, offensive jokes” and called the backlash “pearl clutching”.

The scandal did have repercussions for the Young Republicans, and some senior Republican leaders did condemn the messages. But the fact Vance and others could even think to minimise such vile language speaks to the way far-right politics and sentiments have been normalised today – especially by some in the mainstream media.

As detailed in a book I recently edited, The Far Right and the Media: International Trends and Perspectives, mainstream journalism does not simply cover far-right politics from a critical distance, it also helps define what counts as politically acceptable.

And in many ways, the media is failing in this regard.

Euphemisms and evasion

The first problem has to deal with language itself. When describing the far right, some media outlets reach for softening descriptors such as “populist”, “controversial” or “anti-establishment”, avoiding more accurate terms like “racist” or “authoritarian”.

These linguistic choices are not merely stylistic; they also determine how audiences interpret events and understand what is politically at stake.

Studies of Spanish and Portuguese media have shown, for example, how journalists labelled far-right parties such as VOX and Chega as simply “conservative”, rarely acknowledging their ideological roots in racial nationalism.

In Germany, reporting on the misogynist incel movement has frequently reduced gendered violence to a matter of individual pathology instead of linking it to broader ideological networks of the far right.

In Australia, the mainstream media often treats racialised fears about demographic “threats” as legitimate national concerns.

For example, some commentary has suggested immigration will hurt “Australia’s way of life” or “provoke more internal hostility”. This is frequently framed as a neutral worry about the country’s future.

Yet, this framing overlooks how such claims draw on historical, settler-colonial logic that has cast both First Nations peoples and non-white migrants as populations to be controlled or contained.

When spectacle replaces substance

Sensationalist media coverage of far-right groups can also ensure their views are amplified. And far-right actors have long understood how to manipulate the media by provoking outrage, knowing such acts guarantee attention.

Under commercial pressure, news outlets often take the bait, producing stories that inflate the significance of far-right agitation while neglecting the deeper social and economic conditions that sustain discriminatory politics.

This, in turn, helps to normalise hateful rhetoric.

Research from Loughborough University illustrated this dynamic during the United Kingdom’s 2024 election campaign. Far-right Reform leader Nigel Farage was the third-most-covered political figure, despite his party’s limited electoral prospects. The volume of attention far outweighed his political relevance at the time.

Reform UK was also the only political party to feature in more “good” news than “bad”, the study found.

In this way, visibility achieved through sensationalism can function as a proxy for legitimacy.

False balance and the illusion of neutrality

This emphasis on spectacle over substance is compounded by another long-standing journalistic practice: the performance of balance.

Some media outlets feel compelled to bring balance to stories about those with far-right views by including their denials, justifications or attempts to distract.

In the US, this is the product of decades of industry restructuring. The Federal Communications Commission’s repeal of the Fairness Doctrine in 1987 was formative in this transformation. Not only did it create a path for explicitly partisan media outlets to emerge, it also encouraged mainstream organisations to perform neutrality through superficial “both-sides” reporting.

The coverage of the Young Republicans clearly illustrates this. Rather than examining how racism became embedded within party youth networks, some reporting drew parallels with violent text messages sent by a Democratic candidate for attorney-general in Virginia.

Other media outlets quoted White House officials seeking to divert attention to the Democrats in the same way – in the name of balance.

This reduced the Young Republicans scandal to just another partisan talking point, instead of a moment of reckoning.

Rethinking the media’s role

Through these ways of framing stories, media institutions have functioned as active, if often ambivalent, participants in shaping far-right visibility, rather than as passive conduits exploited by opportunistic actors.

What’s necessary – and entirely possible – is coverage that accurately describes far-right ideology for what it is, situates it within historical and social contexts, and resists the privileging of spectacle over substance.

Only by understanding these dynamics can news organisations begin to counter the forces they so often, however unintentionally, help to sustain.

Imogen Richards, Senior Lecturer in Criminology, Deakin University

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

From next Wednesday, thousands of young Australians under 16 will lose access to their accounts across ten social media platforms, as the teen social media ban takes effect.

What do young people think about it? Our team of 14 leading researchers from around the country interviewed 86 young people from around Australia, aged between 12 and 15, to find out.

Young people’s voices matter

The social media ban, which was legislated 12 months ago, has attracted considerable media coverage and controversy.

But largely missing from these conversations has been the voices of young people themselves.

This is a problem, because research shows that including young people’s voices is best practice for developing policy that upholds their rights, and allows them to flourish in a digital world.

There’s also evidence that when it comes to public policy concerning young people and their use of technology, discussion often slips into a familiar pattern of moral panic. This view frames young people as vulnerable and in need of protection, which can lead to sweeping “fixes” without strong evidence of effectiveness.

‘My parents don’t really understand’

Our new research, published today, centres the voices of young people.

We asked 86 12–15-year-olds from around Australia what they think about the social media ban and the kinds of discussions they’ve had about it. We also asked them how they use social media, what they like and don’t like about it, and what they think can be done to make it better for them.

Some young people we spoke to didn’t use social media, some used it every now and then, and others were highly active users. But they felt conversations about the ban treated them all the same and failed to acknowledge the diverse ways they use social media.

Many also said they felt adults misunderstand their experiences. As one 13-year-old boy told us:

I think my parents don’t really understand, like they only understand the bad part not the good side to it.

Young people acknowledge that others may have different experiences to them, but they feel adults focus too much on risks, and not enough on the ways social media can be useful.

Many told us they use social media to learn, stay informed, and develop skills. As one 15-year-old girl said, it also helps with hobbies.

Even just how to like do something or like how to make something, I’ll turn to social media for it.

Social media also helps young people find communities and make connections. It is where they find their people.

For some, it offers the representation and understanding they don’t get offline. It is a space to explore their identity, feel affirmed, and experience a sense of belonging they cannot always access in their everyday lives.

One 12-year-old girl told us:

The ability to find new interests and find community with people. This is quite important to me. I don’t have that many queer or neurodivergent friends – some of my favourite creators are queer.

Their social media lives are complex and they feel like the ban is an overly simplistic response to the issues and challenges they face when using social media. As one 12-year-old boy put it:

Banning [social media] fully just straight up makes it a lot harder than finding a solution to the problem […] it’s like taking the easy solution.

So what do they think can be done to make social media a better place for them?

Nuanced restrictions and better education

Young people are not naive about risks. But most don’t think a one-size-fits-all age restriction is the solution. A 14-year-old boy captured the views of many who would rather see platforms crack down on inappropriate and low-quality content:

I think instead of doing like a kids’ version and adult version, there should just be a crackdown on the content, like tighter restrictions and stronger enforcement towards the restrictions.

They also want to see more nuanced restrictions that respond to their different ages, and believe platforms should be doing more to make social media better for young people. As one 13-year-old boy said:

Make the platforms safer because they’re like the person who can have the biggest impact.

Young people also want to see more – and crucially, better – education about using social media that takes a more holistic approach and considers the positives that using social media can have for young people. One 15-year-old boy said:

I’d rather [the government] just like implement more media literacy programs instead of just banning [social media] altogether, because it just makes things a lot more complicated in the long run.

As the teen social media ban edges closer and platforms start to implement the legislation, there are practical things children and teens can do to prepare for these changes.

Kim Osman, Senior Research Associate, Digital Media Research Centre, Queensland University of Technology; Lynrose Jane Genon, PhD Candidate, Digital Media Research Centre, Queensland University of Technology, and Michael Dezuanni, Professor, Digital Media Research Centre, Queensland University of Technology

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

Microsoft just released its latest small language model that can operate directly on the user’s computer. If you haven’t followed the AI industry closely, you might be asking: what exactly is a small language model (SLM)?

As AI becomes increasingly central to how we work, learn and solve problems, understanding the different types of AI models has never been more important. Large language models (LLMs) such as ChatGPT, Claude, Gemini and others are in widespread use. But small ones are increasingly important, too.

Let’s explore what makes SLMs and LLMs different – and how to choose the right one for your situation.

Firstly, what is a language model?

You can think of language models as incredibly sophisticated pattern-recognition systems that have learned from vast amounts of text.

They can understand questions, generate responses, translate languages, write content, and perform countless other language-related tasks.

The key difference between small and large models lies in their scope, capability and resource requirements.

Small language models are like specialised tools in a toolbox, each designed to do specific jobs extremely well. They typically contain millions to tens of millions of parameters (these are the model’s learned knowledge points).

Large language models, on the other hand, are like having an entire workshop at your disposal – versatile and capable of handling almost any challenge you throw at them, with billions or even trillions of parameters.

What can LLMs do?

Large language models represent the current pinnacle of AI language capabilities. These are the models making headlines for their ability to “write” poetry, debug complex code, engage in conversation, and even help with scientific research.

When you interact with advanced AI assistants such as ChatGPT, Gemini, Copilot or Claude, you’re experiencing the power of LLMs.

The primary strength of LLMs is their versatility. They can handle open-ended conversations, switching seamlessly from discussing marketing strategies to explaining scientific concepts to creative writing. This makes them invaluable for businesses that need AI to handle diverse, unpredictable tasks.

A consulting firm, for instance, might use an LLM to analyse market trends, generate comprehensive reports, translate technical documents, and assist with strategic planning – all with the same model.

LLMs excel at tasks requiring nuanced understanding and complex reasoning. They can interpret context and subtle implications, and generate responses that consider multiple factors simultaneously.

If you need AI to review legal contracts, synthesise information from multiple sources, or engage in creative problem-solving, you need the sophisticated capabilities of an LLM.

These models are also excellent at generalising. Train them on diverse data, and they can extrapolate knowledge to handle scenarios they’ve never explicitly encountered.

However, LLMs require significant computational power and usually run in the cloud, rather than on your own device or computer. In turn, this translates to high operational costs. If you’re processing thousands of requests daily, these costs can add up quickly.

When less is more: SLMs

In contrast to LLMs, small language models excel at specific tasks. They’re fast, efficient and affordable.

Take a library’s book recommendation system. An SLM can learn the library’s catalogue. It “understands” genres, authors and reading levels so it can make great recommendations. Because it’s so small, it doesn’t need expensive computers to run.

SLMs are easy to fine-tune. A language learning app can teach an SLM about common grammar mistakes. A medical clinic can train one to understand appointment scheduling. The model becomes an expert in exactly what you need.

SLMs are faster than LLMs, too – they can deliver answers in milliseconds, rather than seconds. This difference may seem small, but it’s noticeable in applications such as grammar checkers or translation apps, which can’t keep users waiting.

Costs are much smaller, too. Small language models are like LED bulbs – efficient and affordable. Large language models are like stadium lights – powerful but expensive.

Schools, non-profits and small businesses can use SLMs for specific tasks without breaking the bank. For example, Microsoft’s Phi-3 small language models are helping power an agricultural information platform in India to provide services to farmers even in remote places with limited internet.

SLMs are also great for constrained systems such as self-driving cars or satellites that have limited processing power, minimal energy budgets, and no reliable cloud connection. LLMs simply can’t run in these environments. But an SLM, with its smaller footprint, can fit onboard.

Both types of models have their place

What’s better – a minivan or a sports car? A downtown studio apartment or a large house in the suburbs? The answer, of course, is that it depends on your needs and your resources.

The landscape of AI models is rapidly evolving, and the line between small and large models is becoming increasingly nuanced. We’re seeing hybrid approaches where businesses use SLMs for routine tasks and escalate to LLMs for complex queries. This approach optimises both cost and performance.

The choice between small and large language models isn’t about which is objectively better – it’s about which better serves your specific needs.

SLMs offer efficiency, speed and cost-effectiveness for focused applications, making them ideal for businesses with specific use cases and resource constraints.

LLMs provide unmatched versatility and sophistication for complex, varied tasks, justifying their higher resource requirements when a highly capable AI is needed.

Lin Tian, Research Fellow, Data Science Institute, University of Technology Sydney and Marian-Andrei Rizoiu, Associate Professor in Behavioral Data Science, University of Technology Sydney

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