Tyler Robinson, the 22-year-old Utah man suspected of having fatally shot right-wing activist Charlie Kirk, is reportedly not cooperating with authorities.

Robinson was apprehended after a more than two-day manhunt and is being held without bail at the Utah County Jail.

While a motive for the shooting has yet to be established, Utah Governor Spencer Cox has highlighted Robinson’s links to gaming and the “dark internet”.

Bullet casings found at the scene were inscribed with various messages evoking gaming subcultures. One of the quotes – “Notices bulges, OwO what’s this” – can be linked to the furry community, known for role-playing using animal avatars.

Another message – “Hey, fascist! Catch! ↑ → ↓↓↓” – features arrow symbols associated with an action that allows players to drop bombs on their foes in Helldiver 2, a game in which players play as fascists fighting enemy forces.

One casing reads “O Bella ciao, Bella ciao, Bella ciao, Ciao, ciao!”, words from an Italian anti-Mussolini protest song, which also appears in the shooter game Far Cry 6. Yet another is a homophobic jibe: “if you read this you are gay LMAO”.

If Robinson does turn out to be a shooter radicalised through online gaming spaces, he would not be the first. Previous terrorist shootings at Christchurch (New Zealand), Halle (Germany), Bærum (Norway), and the US cities of Buffalo, El Paso and Poway were all carried out by radicalised young men who embraced online conspiracies and violent video games.

In each of these cases, the shooter attempted (and in all but the Poway shooting, succeeded) to live stream the atrocities, as though emulating a first-person shooter game.

A growing online threat

The global video game market is enormous, with an estimated value of almost US$300 billion (about A$450 billion) in 2024. Of the more than three billion gamers, the largest percentage is made up of young adults aged 18–34.

Many of these are vulnerable young men. And extremist activists have long recognised this group as a demographic ripe for radicalisation.

As early as 2002, American neo-Nazi leader Matt Hale advised his followers “if we can influence video games and entertainment, it will make people understand we are their friends and neighbours”.

Since then, far-right groups have produced ethnonationalist-themed games, such as “Ethnic Cleansing” and “ZOG’s Nightmare”, in which players defend the “white race” against Islamists, immigrants, LGBTQIA+ people, Jews and more.

Studying radicalisation in gamer circles

For many, the Kirk shooting has resurfaced the perennial question about the link (or lack thereof) between playing violent video games and real-world violence.

But while this is an important line of inquiry, the evidence suggests most radicalisation takes place not through playing video games themselves, but through gaming platform communication channels.

In 2020, my colleagues and I studied an extraordinary data dump of more than nine million posts from the gaming platform Steam to understand this process.

We found evidence of radicalisation occurring through communication channels, such as team voice channels. Here, players establish connections with one another, and can leverage these connections for political recruitment.

The radicalisation of vulnerable users is not instantaneous. Once extremists have connected with potential targets, they invite them into platforms such as Discord or private chat rooms. These spaces allow for meme and image sharing, as well as ongoing voice and video conversations.

Skilful recruiters will play to a target’s specific grievances. These may be personal, psycho-sexual (such as being unable to gain love or approval), or related to divisive issues such as employment, housing or gender roles.

The recruit is initiated into a fast-changing set of cynical in-jokes and in-group terms. These may include mocking self-designations, such as the Pepe the Frog meme, used by the far-right to ironically embrace their ugly “political incorrectness”. They also use derogatory terms for “enemies”, such as “woke”, “social justice warriors”, “soyboys”, “fascists” and “cultural Marxists”.

Gradually, the new recruit becomes accustomed to the casual denigration and dehumanisation of the “enemies”.

Dark and sarcastic humour allow for plausible deniability while still spreading hate. As such, humour acts an on-ramp to slowly introduce new recruits to the conspiratorial and violent ideologies that lie at the heart of terrorist shootings.

Generally, these ideologies claim the world is run by nefarious and super-powerful plutocrats/Jews/liberals/communists/elites, who can only be stopped through extreme measures.

It then becomes a question of resolve. Who among the group is willing to do what the ideology suggests is necessary?

What can be done?

The Australian Federal Police, as well as the Australian parliament, has recognised the threat of violence as a result of radicalisation through online gaming. Clearly, it’s something we can’t be complacent about.

Social isolation and mental illness, which are sadly as widespread in Australia as they are elsewhere, are some of the factors online extremists try to exploit when luring vulnerable individuals.

At the same time, social media algorithms function to shunt users into ever more sensational content. This is something online extremists have benefited from, and learned to exploit.

There is a growing number of organisations devoted to trying to prevent online radicalisation through gaming platforms. Many of these have resources for concerned parents, teachers and care givers.

Ultimately, in an increasingly online world, the best way to keep young people safe from online radicalisation is to keep having constructive offline conversations about their virtual experiences, and the people they might meet in the process.

Matthew Sharpe, Associate Professor in Philosophy, Australian Catholic University

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

Large pictorial warning labels on smoked tobacco products typically feature confronting images of the harmful health outcomes of smoking.

Pictures of diseased lungs, gangrene and mouth decay aim to elicit strong emotional responses that reduce the appeal and acceptability of smoking, particularly among young people.

Warning labels also aim to increase knowledge of the many risks smoking poses. Plain packaging increases the attention paid to warning labels and reduces pack appeal, brand loyalty and product perceptions.

However, like any marketing campaign, warning labels on tobacco products need regular updating so they continue to attract attention and communicate the latest research evidence. Maintaining the same images risks “wear-out”, when people lose interest in campaign images and messages, or counter argue these.

Our recent work found existing tobacco warnings have lost impact. Study participants had created cognitive defences and exempted themselves from the risks shown.

While some still found images of diseased body parts shocking, others did not consider the illustrated risks personally relevant and thought warnings lacked credibility.

Our findings raise the important question of how we can make on-pack warnings more impactful and effective. Our new study addresses this question by exploring the lived experiences of people who smoke.

Creating more effective warnings

Many find smoking imposes a heavy financial burden on them, and others worry about the impact smoking has on whānau (families).

Others resent the hold nicotine addiction has over them and feel concerned that young people may want to adopt the behaviour they see modelled by adults who smoke.

Working with a graphic artist, we developed images and messages to represent the ideas we had heard.

After extensive review with people who smoke, we identified three potential warning themes for final testing: the cost of smoking, smoking’s impact on family, and the health risks presented in a more empathetic way (by featuring people rather than diseased body parts).

Using a choice study, we examined how well warnings representing these themes prompted thoughts of quitting compared to a novel graphic health warning showing a mouth cancer.

We found two different groups among our sample of people who smoke: one responded more strongly to warning labels emphasising the cost of smoking and its effect on families than to the graphic warning we used as a control; the other group reacted more strongly to an empathetic health warning than to the control.

5 ways to improve on-pack warnings

1. We need warnings that reflect people’s experiences of smoking, recognise smoking’s various harms, and understand that people who smoke are not a homogenous group. While most people who smoke regret smoking and hope to quit, they are at different life stages, have different backgrounds and interests, and respond to different stimuli.

For example, the cost-of-living crisis means warnings reinforcing the cost of smoking, the opportunities forgone and the impacts on others may be more motivating for some people than graphic health warnings.

2. We should think more creatively about the health harm from smoking. We found images of children losing a parent to an illness caused by smoking created strong emotional connections, as did images of adults smoking near children.

This approach, which illustrates how smoking causes emotional and physical harm to others, was at least as effective as the graphic mouth cancer image we used as a control.

3. We should consider the impact of warnings on emotions. Early graphic warnings aimed to arouse fear, in the belief it would galvanise attempts to quit. However, people who smoke also experience regret and shame, which may be more motivating than fear.

4. We need to balance negative emotions, which may stigmatise people and lead them to feel powerless, by introducing pack inserts with positive messages. Our work found that offering helpful advice and outlining the benefits of quitting inspired participants and could support attempts to quit.

5. We need to refresh and rotate warnings much more often. We suggest new warnings should be introduced every six months and that no warning should run for longer than a year.

On-pack pictorial warnings are a proven best-practice approach to encouraging smoking cessation. However, failure to introduce new and more diverse warnings has compromised the impact these have.

Given people who smoke consume, on average, ten cigarettes a day, on-pack warnings have high potential exposure. We should be making this measure as effective as possible and embed it within a comprehensive strategy that will reduce tobacco’s addictiveness, appeal and accessibility.

Janet Hoek, Professor in Public Health, University of Otago; Andrew Waa, Associate Professor in Public Health, University of Otago; Lani Teddy, Research Fellow in Public Health, University of Otago, and Philip Gendall, Senior Research Fellow in Marketing, University of Otago

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

Just for a moment this week, Larry Ellison, co-founder of US cloud computing company Oracle, became the world’s richest person.

The octogenarian tech titan briefly overtook Elon Musk after Oracle’s share price rocketed 43% in a day, adding about US$100 billion (A$150 billion) to his wealth.

The reason? Oracle inked a deal to provide artificial intelligence (AI) giant OpenAI with US$300 billion (A$450 billion) in computing power over five years.

While Ellison’s moment in the spotlight was fleeting, it also illuminated something far more significant: AI has created extraordinary levels of concentration in global financial markets.

This raises an uncomfortable question not only for seasoned investors – but also for everyday Australians who hold shares in AI companies via their superannuation. Just how exposed are even our supposedly “safe”, “diversified” investments to the AI boom?

The man who built the internet’s memory

As billionaires go, Ellison isn’t as much of a household name as Tesla and SpaceX’s Musk or Amazon’s Jeff Bezos. But he’s been building wealth from enterprise technology for nearly five decades.

Ellison co-founded Oracle in 1977, transforming it into one of the world’s largest database software companies. For decades, Oracle provided the unglamorous but essential plumbing that kept many corporate systems running.

The AI revolution changed everything. Oracle’s cloud computing infrastructure, which helps companies store and process vast amounts of data, became critical infrastructure for the AI boom.

Every time a company wants to train large language models or run machine learning algorithms, they need huge amounts of computing power and data storage. That’s precisely where Oracle excels.

When Oracle reported stronger-than-expected quarterly earnings this week, driven largely by soaring AI demand, its share price spiked.

That response wasn’t just about Oracle’s business fundamentals. It was about the entire AI ecosystem that has been reshaping global markets since ChatGPT’s public debut in late 2022.

The great AI concentration

Oracle’s story is part of a much larger phenomenon reshaping global markets. The so-called “Magnificent Seven” tech stocks – Apple, Microsoft, Alphabet, Amazon, Meta, Tesla and Nvidia – now control an unprecedented share of major stock indices.

Year-to-date in 2025, these seven companies have come to represent approximately 39% of the US S&P500’s total value. For the tech-heavy NASDAQ100, the figure is a whopping 74%.

This means if you invest in an exchange-traded fund that tracks the S&P500 index, often considered the gold standard of diversified investing, you’re making an increasingly concentrated bet on AI, whether you realise it or not.

Are we in an AI ‘bubble’?

This level of concentration has not been seen since the late 1990s. Back then, investors were swept up in “dot-com mania”, driving technology stock prices to unsustainable levels.

When reality finally hit in March 2000, the tech-heavy Nasdaq crashed 77% over two years, wiping out trillions in wealth.

Today’s AI concentration raises some similar red flags. Nvidia, which controls an estimated 90% of the AI chip market, currently trades at more than 30 times expected earnings. This is expensive for any stock, let alone one carrying the hopes of an entire technological revolution.

Yet, unlike the dot-com era, today’s AI leaders are profitable companies with real revenue streams. Microsoft, Apple and Google aren’t cash-burning startups. They are established giants, using AI to enhance existing businesses while generating substantial profits.

This makes the current situation more complicated than a simple “bubble” comparison. The academic literature on market bubbles suggests genuine technological innovation often coincides with speculative excess.

The question isn’t whether AI is transformative; it clearly is. Rather, the question is whether current valuations reflect realistic expectations about future profitability.

Hidden exposure for many Australians

For Australians, the AI concentration problem hits remarkably close to home through our superannuation system.

Many balanced super fund options include substantial allocations to international shares, typically 20–30% of their portfolios.

When your super fund buys international shares, it’s often getting heavy exposure to those same AI giants dominating US markets.

The concentration risk extends beyond direct investments in tech companies. Australian mining companies, such as BHP and Fortescue, have become indirect AI players because their copper, lithium and rare earth minerals are essential for AI infrastructure.

Even diversifying away from technology doesn’t fully escape AI-related risks. Research on portfolio concentration shows when major indices become dominated by a few large stocks, the benefits of diversification diminish significantly.

If AI stocks experience a significant correction or crash, it could disproportionately impact Australians’ retirement nest eggs.

A reality check

This situation represents what’s called “systemic concentration risk”. This is a specific form of systemic risk where supposedly diversified investments become correlated through common underlying factors or exposures.

It’s reminiscent of the 2008 financial crisis, when seemingly separate housing markets across different regions all collapsed simultaneously. That was because they were all exposed to subprime mortgages with high risk of default.

This does not mean anyone should panic. But regulators, super fund trustees and individual investors should all be aware of these risks. Diversification only works if returns come from a broad range of companies and industries.

Angel Zhong, Professor of Finance, RMIT University and Jason Tian, Senior Lecturer, Swinburne University of Technology

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