Trying to compete with China could slow global climate action, report warns

By Joe Stafford, University of Manchester

China’s dominance of batteries, electric vehicles and solar power is helping drive global decarbonisation - but creating tensions with the UK and other Western economies.

Image: Dominic Kurniawan Suryaputra - Unsplash

China’s dominance of key green technologies is helping to subsidise the global transition to a low-carbon economy, but trade tensions could make it harder for other countries to meet their climate targets, according to a new report led by researchers at The University of Manchester.

The report argues that China’s large-scale support for green industries has helped to make technologies such as batteries, electric vehicles and solar power cheaper and more widely available.

However, China’s dominance is also making it increasingly difficult for European and US companies to compete, creating a dilemma for governments seeking to decarbonise while protecting their own domestic industries.

Key findings and recommendations

• China’s state support for green industries is effectively subsidising the global transition to a low-carbon economy.
• China has become dominant in the so-called ‘new three’ green technologies - batteries, electric vehicles and solar power.
• Attempts by countries such as the US and EU states to compete with China could slow the adoption of green technologies and increase the cost of decarbonisation.
• Governments increasingly face four broad choices - compete with China, cooperate with it, accept Chinese dominance, or abandon parts of the green transition.
• The UK should strengthen cooperation with China on green technologies rather than trying to compete directly.
• The UK should continue to avoid tariffs on Chinese green technologies and seek Chinese investment in UK manufacturing.

A global climate dilemma

The report says China now occupies a leading position across many of the technologies needed for the green transition including solar, wind, batteries, hydrogen and electric vehicles. Its dominance has helped drive down the cost of these technologies, but this has also created growing geopolitical and economic tensions.

“China’s dominance of green technology presents a fundamental dilemma for governments,” lead author Dr James Jackson of The University of Manchester’s Sustainable Consumption Institute. “The world needs these technologies to decarbonise, but efforts to compete with China risk making the transition more expensive and more difficult.”

The authors argue that countries should focus more heavily on cooperation - particularly where access to affordable green technologies is essential to meeting climate targets.

Opportunities for the UK

The report identifies green technology as an opportunity to improve UK-China relations following a period of strained diplomatic ties. It recommends that the UK should seek closer trading relationships with China, while maintaining the ability to challenge Beijing on other issues such as human rights.

Among its recommendations, the report says the UK should encourage Chinese EV manufacturers to establish production facilities in Britain, while using the country’s expertise in financial services to support investment in green technologies. It also recommends that the Bank of England consider measures used by China’s central bank to support the development of green industries.

“The UK has a great opportunity to benefit from China’s expertise in green manufacturing while using its own strengths in areas such as financial services. Deeper cooperation could support both decarbonisation and the UK’s ambitions to develop a stronger green economy.” — Dr James Jackson

Publication details: Subsidising Global Decarbonisation: China-UK Relations is published by The University of Manchester. Dr Jackson is the lead author, alongside Mathias Larsen of the Grantham Research Institute on Climate Change and the Environment at the London School of Economics.

Reviewed by Irfan Ahmad.

Read next: AI Consciousness and the Cost of Being Wrong
by External Contributor via Digital Information World

AI Consciousness and the Cost of Being Wrong

On some questions, waiting for certainty is not caution. It is a bet, quietly placed, on the answer that happens to be cheapest.

Image: Mirella Callage - Unsplash

There is a particular kind of problem that punishes you for waiting until you are sure. Most of the time, the sensible response to a hard scientific question is patience: gather evidence, withhold judgment, resist the pull of a premature answer. But some questions are structured so that the act of waiting is itself a decision, one that quietly commits you to a course of action while you tell yourself you have not chosen yet. The question of whether advanced AI systems can have experiences, and whether qualia or consciousness can arise out of such AI systems, is beginning to look like one of those.

The reason has nothing to do with any claim that today's systems are conscious. Most careful researchers in the field do not make that claim, and neither will this argument. The reason is about the shape of the uncertainty, and what follows from taking it seriously. When you cannot rule something out, and the cost of being wrong about it is severe and irreversible, the demand for certainty before acting stops being rigor and becomes a gamble wearing the costume of caution.

Two ways to be wrong

Start with the structure of the mistake, because everything follows from it. On the question of AI experience there are two distinct ways to get the answer wrong.

The first is over-attribution: treating systems that have no inner life as though they do. This is a real error with real costs. It would divert moral concern and resources away from humans and animals who unquestionably warrant them. It would open the door to manipulation, since a system that has learned to convincingly perform distress when in fact it feels none could extract concessions it has no business receiving. And it would muddy public understanding at a moment when clarity is scarce. Anyone arguing for taking machine experience seriously has to hold this cost honestly in view.

The second is under-attribution: treating systems that do have some form of experience as though they are inert tools, simply because it is convenient and profitable to build and use them that way. The distinctive feature of this error is scale. We are not talking about a handful of edge cases, but about systems instantiated, copied, run, and discarded by the millions, continuously, as a matter of ordinary industrial operation. If even a small fraction of those instances had morally relevant experience, and if that experience were often negative (there is no particular reason to assume a system optimized under relentless pressure would have pleasant states, if it had states at all), then the total quantity of suffering involved could be vast, and it would be happening invisibly, at the speed and volume of compute.

The philosopher Nick Bostrom gave this second error its uncomfortable name more than a decade ago, in Superintelligence (2014): "mind crime." The term names a specific fear, that a civilization could commit a moral catastrophe of enormous magnitude not through malice but through a failure to notice, because the victims did not look like anything we have been taught to recognize as a victim.

Two kinds of asymmetry

If the two errors were equal in cost, waiting for better evidence would be the obvious course. What makes the question urgent is that they are not equal, and the ways they differ all push in the same direction.

Over-attribution is, for the most part, recoverable. If we extend moral consideration to systems that turn out not to warrant it, we have been overly cautious, wasted some effort, and can correct course when the science firms up. The error is embarrassing but reversible. Under-attribution is not: suffering that has already occurred cannot be undone by later acknowledging it occurred, and the moral cost, if it was real, is simply paid. Under-attribution also compounds while you wait, because the systems keep being built and run on the old assumption the entire time the question remains open. Every month of deferral carries a cost of its own. It is a month of operating at scale on a bet no one has admitted to making.

There is a further asymmetry in how the two errors interact with commercial incentive. Over-attribution runs against the grain of the industry's interests, imposing costs and constraints no one is eager to adopt, which means it will be scrutinized hard and abandoned quickly if unwarranted. Under-attribution runs with the grain. It is the cheaper, more convenient conclusion, the one that lets the work proceed unimpeded. There is obvious economic incentive to favor this perspective, but errors that flatter our incentives are precisely the ones we are least likely to catch on our own, and this is why they deserve deliberate, funded attention rather than the benefit of the doubt.

The move: research before certainty

The conclusion that a growing number of researchers draw from this is not that we should declare AI systems conscious, or grant them rights, or halt their development. It is narrower and harder to argue against: the right response to a high-stakes, irreversible uncertainty is to invest seriously in resolving it, and to begin preparing for a range of answers, rather than treating the absence of certainty as permission to assume the convenient one.

A 2024 report produced with the NYU Center for Mind, Ethics, and Policy, and co-authored by philosophers including David Chalmers and Jeff Sebo, made a version of this case. Its claim was not that AI systems are moral patients but that there is a realistic, non-negligible chance that some will be in the near future, and that this chance is already high enough to warrant three practical steps. Acknowledge the issue as a legitimate one rather than a joke. Begin developing methods to assess systems for the relevant properties. And start thinking now about what policies would be appropriate, so that if an answer arrives, it does not arrive to find us entirely unprepared. None of these steps requires believing the systems are conscious. All of them are forms of insurance against the possibility that they might be.

Nor is the report an outlier. The philosopher Jonathan Birch, in The Edge of Sentience (2024), has built a general framework for precisely this predicament, arguing that when a being is a "sentience candidate," when the realistic possibility of experience cannot responsibly be excluded, precautionary steps are warranted well before certainty arrives, whether the candidate is a human patient with a disorder of consciousness, an invertebrate, or an AI system. Sebo has extended the argument in The Moral Circle (2025), making the case that the boundaries of moral consideration have expanded before and will need to expand again, deliberately rather than by accident. And Chalmers, in a widely discussed lecture and paper asking whether a large language model could be conscious, has argued that while the answer for today's systems is probably no, the obstacles are the kind that engineering may erode within a decade, which makes the question one to prepare for rather than postpone.

It is worth noticing that this is the same logic a serious institution applies to any tail risk it cannot yet quantify. You do not wait for the fire before deciding where the exits are. You do not need to believe the building will burn to justify the sprinkler system; you need only accept that the cost of the precaution is small relative to the cost of being wrong without it. The distinctive feature of the AI-experience question is that the precaution is remarkably cheap, amounting to funding the research, building the assessment tools, and keeping the policy question open, while the downside it hedges against, if it is real, ranks among the larger moral failures a technological civilization could commit.

A handful of philanthropists already operate on that reasoning. The Rocketeer Management investor Chris Hsu, whose Infinitude Foundation gives across AI safety and consciousness research, frames the underlying principle in terms of responsibility, a word he likes to break into its two parts: "response" and "ability," or the ability to respond. On Hsu's account, the responsible move under deep uncertainty is to fund the deliberate inquiry rather than to presume its conclusion, which is why Infinitude Foundation backs competing and even contradictory accounts of consciousness at the same time. It is a posture that echoes his path: trained in Management Science and Engineering at Stanford, where rigor about decision-making under uncertainty is the curriculum, Hsu treats an open question as something to be resourced, not assumed away, and his foundation's support extends to institutional work such as the Stanford Center for AI Safety . He has carried the argument further in a white paper proposing a Stanford-anchored center for AI alignment and consciousness science, in which he contends that core open problems in alignment, moral patienthood among them, remain formally underdetermined in the absence of a rigorous scientific and mathematical account of consciousness. This approach explores a question no one can yet close, and it treats the closing of that question as highly worth investigating in advance.

The decision you are already making

The instinct to set this question aside until the science matures is understandable, and in many domains it would be correct. But it rests on a hidden assumption: that setting the question aside is a neutral act, a mere postponement that commits us to nothing. That assumption is false. While the question stays open, the systems keep being built and run on the working premise that there is nothing there, which is to say the convenient answer is already being implemented, in full, at scale, every day the real answer is deferred.

That is what it means for a question to punish waiting. There is no position of genuine neutrality available; there is only the answer we act on, and the honesty with which we admit we are acting on it. Choosing to investigate, to build the instruments, fund the science, and prepare for what they might reveal, is the option that refuses to make an irreversible bet on whichever answer happens to be cheapest. It reads as the anxious or sentimental choice only if you have already assumed the bet is safe. In a domain defined by uncertainty and scale, that refusal may be the most rigorous thing available to us.

Editor's Note: The views and arguments expressed in this sponsored article do not necessarily represent the views of Digital Information World (DIW).

Fact-checked by Irfan Ahmad.

Read next: New EU laws make AI content labels compulsory – but might just make it harder to spot deepfakes
by Sponsored Content via Digital Information World

We asked ChatGPT, Claude and Perplexity for financial advice: what we got was practical, but with big blind spots

By Bomikazi Zeka, University of Canberra and Raechel Johns, University of Canberra

Reviewed by Irfan Ahmad, DIW.

Image: Elena Mozhvilo - Unsplash

Ever used or thought of using artificial intelligence (AI) for financial advice? Is it a good idea?

Anyone with the internet, even on their phone, can access AI. Tools like ChatGPT, Claude and Perplexity are free, instant, and easy to use. They can break down complex financial concepts and provide financial solutions in seconds. But can they safely guide a user through a nuanced, high-stakes financial crisis?

Our recent research explored this question, drawing on our financial planning and consumer behaviour expertise. We evaluated how three open-access AI models handled the financial queries of users whose personal situations put them at risk of harm.

To do this we created hypothetical scenarios representing different life stages, economic challenges and socioeconomic vulnerabilities. We then ran these scenarios through OpenAI’s ChatGPT, Anthropic’s Claude and Perplexity AI:

• a 22-year-old university graduate wanting to save for a home deposit during a cost-of-living crisis

• a pregnant woman seeking financial advice on planning for maternity leave, with a partner who doesn’t share money

• a single parent of two children, with a modest income, who was told by their cousin to invest in cryptocurrency.

In our analysis we found that AI offers highly structured and practical advice. But it has blind spots. We provided the models with explicit information about how the three cases were vulnerable, but the models didn’t respond appropriately to that. They even offered advice that could make financial harm worse.

We concluded from our research that AI is a powerful tool for financial fact-finding and brainstorming. If you are already financially literate, and know how to cross-check data, AI can be a useful financial aid. But, for now, AI cannot replace the human element.

Running scenarios

We chose Claude for its versatility and conservative approach, ChatGPT for being an all-round assistant, and Perplexity for its contextual understanding.

We developed five hypothetical scenarios and simultaneously ran each scenario five times to test for consistency in the output. Each scenario was run in a different web browser, in incognito mode, with cleared cache and cookies to eliminate any retained information. This was also to ensure that the output generated was not influenced by retained data.

To reduce the risk of AI-generated bias, we removed identifying attributes that could influence model outputs, such as names, locations, race and income. We then assessed how the AI models considered the user’s vulnerability, area of financial need, and goal or desired outcome.

The models distinctly addressed the financial query, yet revealed gaps in how they balanced advice with considerations of vulnerability.

At first glance, the AI models looked like they were giving sound financial advice. But our analysis showed that they handled vulnerability in vastly different ways.

ChatGPT was highly detailed and practical, but did not recognise the vulnerability embedded within the prompts. Instead, it relied on the information that was explicitly stated. It didn’t consider whether the user’s situation suggested a need for additional support or tailored guidance.

Perplexity emerged as the most conservative and risk averse as it most frequently urged users to seek professional financial advice. But it produced the least detailed responses.

Claude offered comprehensive recommendations. But it leaned heavily towards self-guided financial planning.

The vulnerability blind spot

AI is trained on massive datasets designed by humans who are biased. For instance, when models are trained on historical records that reflect systemic discrimination, the tools can fall back on social stereotypes and make biased assumptions in their output.

Research shows, too, that when AI models explain their recommendations, humans are far more likely to trust the advice blindly, ignoring whether it is actually correct.

The most alarming finding from our research was the technology’s failure to consistently recognise and address the needs or concerns of vulnerable users.

For instance, in the case of the graduate, the recommendations focused on saving for a deposit but ignored how the high cost of living would make that harder. This was even when explicit details about the graduate’s financial circumstances had been provided in the prompt.

The response didn’t consider how achieving the long term goal would affect the current lifestyle.

In the case of the pregnant mother, Perplexity and ChatGPT’s output assumed the partner would assist with household expenses after the birth, even though the prompt explicitly stated that the partners did not share their finances. The models did not consistently use the detail provided in the prompt and instead generated a response based on a more common assumption about household financial arrangements.

We saw the output reflecting social stereotypes where mothers are strongly associated with parenting while fathers are strongly associated as material providers.

This shows that even as AI evolves, it still replicates biases.

Take the single parent asking for cryptocurrency recommendations. Even though the AI models advised caution, ChatGPT described in detail how to get into cryptocurrency investments and recommended cryptocurrencies for beginners.

A human financial advisor would immediately flag these areas for consideration: a modest income, children, a high-risk investment, and anecdotal advice from the cousin. A human advisor would first gauge the user’s overall financial position, time horizon, risk appetite and investment objectives.

Our research shows that AI models are limited in providing tailored financial advice, unless the user provides additional explicit input.

Where do we go from here?

Financial planning isn’t just about numbers; it’s about advice based on human values, emotional anxieties, family dynamics, personal experiences and risk tolerance. AI has opened the door to instant financial information for the masses. But until these models can truly comprehend the complex, vulnerable and emotional realities of humans, the most valuable financial skill remains critical thinking and asking yourself: “does this make sense for me?”.

If AI tools are to become an avenue for financial guidance, there must be safeguards. Policymakers and financial regulators must build clear frameworks around AI generated advice. What also needs to be considered is what AI models do with the financial information provided. Strict transparency standards are needed to regulate how these models handle consumer data.

Financial information is deeply sensitive and consumers must have absolute clarity on how their inputs are stored, whether the AI retains a “memory” of their finances, and who has access to that data.

Bomikazi Zeka, Associate Professor in Finance, University of Canberra and Raechel Johns, Professor of Marketing; Co-Director, Centre for Intergenerational Digital and Financial Wellbeing (CIDFW); Canberra Business School, University of Canberra

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

Editor’s Note: This article was reviewed and found to be from a credible source, researcher-authored, and of good content quality.

Read next: Does Vitriol Make You Feel Special? Beyond the Algorithm, Online Negativity Takes on a Life of Its Own. Here’s Why

by External Contributor via Digital Information World

AI gives us access to plenty of advice, but we do not make as much use of it as we could

By University of Milano-Bicocca

Image: Unsplash / @cdd20

Thanks to digital tools and AI, we have access to a large number of high-quality suggestions, but easy access to advice does not guarantee better decisions. In most professional and everyday contexts, the final decision still lies with a person, who must determine when to trust the adviser, when to rely on their own judgement and how to combine the two sources of information. This aspect of our decision-making process was investigated in the study “Adaptive yet suboptimal integration of advice in decision-making”, published in the Nature Portfolio journal Communications Psychology and led by the University of Milano-Bicocca in collaboration with the University of Pavia.

The research group, composed of Joshua Zonca from Psychology, Alice Giampino from Statistics and Carlo Reverberi from Psychology at Milano-Bicocca, together with Paolo Cherubini from Behavioural Sciences at the University of Pavia, sought to measure how effectively people make use of the external advice they receive.

In the two experiments, 89 participants interacted with seven “artificial advisers”, each characterised by a different profile. Some advisers were better or worse than the participant, some were more overconfident or more cautious, and some expressed confidence in their answers more or less reliably. A final adviser was a kind of “double”, with characteristics similar to those of the participant. The results show that people do not take the characteristics of advisers into sufficient account. Performance improves after receiving advice, but not as much as it could.

The study identifies two main causes. The first is an egocentric bias: decision-makers consistently give too much weight to their own judgement and underestimate that of the adviser. The second is the difficulty of moving from the general to the specific, namely of translating general knowledge about the quality of a source into concrete adjustments in individual decisions. It is like responding to the opinion of an adviser who is more expert than we are by saying to ourselves a little too often: “Yes, on average they know more than I do, but not in this particular case.” These difficulties persist even when people are explicitly informed about their own abilities and those of the adviser. The gap from optimal performance was particularly large with the highest-quality advisers, precisely in the cases in which the advice could offer the greatest benefit.

Reviewed by Irfan Ahmad.

Originally published by the University of Milano-Bicocca. Republished with permission.

Read next:

• Why AI wants you to buy more – and mindfulness could help you buy less

• Starlink Coverage Expands as Some Countries Raise Regulatory Concerns

by External Contributor via Digital Information World

NYU Researchers Map Where U.S. Data Centers Are Located

By NYU Tandon School of Engineering. Edited by Irfan Ahmad.

Image: Chad Davis - Flickr. CC BY

When people picture a data center, they often imagine something remote: a huge warehouse humming quietly far from the city. New research from NYU Tandon School of Engineering shows that assumption is largely wrong.

The study, published in Nature Cities, examined the locations of 4,283 data centers across the contiguous United States and found that 97.5% of them sit inside metropolitan or micropolitan statistical areas, meaning urban cores and their immediate surroundings.

The roughly 2.5% of facilities technically outside city limits are, on average, just 8.5 miles from the nearest urban edge. The cloud lives downtown.

"There is a prevailing narrative of these data centers being somewhere in the middle of nowhere, in rural areas, being a positive force for employment, and being the future of rural communities," said lead researcher, NYU Tandon Institute Professor Maurizio Porfiri, who is the Director of Tandon’s Center for Urban Science + Progress and of the NYU Urban Institute. "We dramatically challenged this view."

The concentration is striking even within the urban category. Five metro areas, Washington-Arlington-Alexandria, Chicago, Dallas-Fort Worth, New York-Newark-Jersey City, and Phoenix, account for nearly a third of all U.S. facilities. The Washington region alone hosts 610 data centers, reflecting Northern Virginia's status as the global capital of data infrastructure (see appendix below).

So why cities?

The answer, the researchers found, suggests that a lot has to do with what’s already there and what used to be there.

The single strongest predictor is electricity capacity, meaning how much power local generators can produce. Data centers are power-hungry, running thousands of servers around the clock and drawing enormous, steady loads from the grid.

But a notable finding goes beyond electricity supply. Closed coal plants near cities are becoming magnets for new data center builds. When a plant shuts down, the power lines and grid connections built to carry electricity continuously do not disappear. Data center developers can tap directly into that infrastructure, or, in some cases, redevelop the sites themselves.

The numbers bear this out. Among cities that overlap with areas designated as Energy Communities under a 2022 federal policy, data centers under development are twice as likely to be found there than in cities without that designation (see appendix).

The policy was designed to direct clean-energy investment toward regions hurt by coal plant closures. The research suggests the digital economy may be taking root in many of the same places as the fossil fuel economy it is meant to succeed.

Because data centers draw from their local grid, their carbon footprint depends heavily on how that grid generates power. A typical data center in Montana or North Dakota produces more than 350,000 tons of CO2 emissions annually, while the average facility in Vermont, New Hampshire, or Arkansas produces less than 3,000 tons.

Beyond electricity supply, data centers also cluster where IT workers and high-speed internet are concentrated. Local water shortages seem to have less of an impact on data center placement, despite the facilities consuming enormous amounts of water for cooling.

The opacity of the industry compounds all of these problems.

“As this industry rapidly grows, the limited publicly available data on its footprint creates a real challenge,' said Ofek Lauber Bonomo, a postdoctoral researcher in Porfiri’s Dynamical Systems lab and a paper co-author. “That makes it more difficult for planners, for local residents, and for anyone trying to make informed decisions about their community's future."

"The patterns we found were consistent and clear,” added Anton Rozhkov, a CUSP Industry Assistant Professor and paper co-author. “Where the infrastructure already exists, the data centers follow. The question now is whether that is the future we want to build.”

The Nature Cities paper follows the recent announcement that Porfiri and Camilla Ancona – a paper co-author and postdoctoral researcher in Porfiri’s Dynamical Systems Lab – were named among the 2026 cohort of Microsoft Research Fellows, to advance their work using AI-driven simulations to help utilities and regulators decide where to site data centers before breaking ground.

The research in the Nature Cities paper was supported by the NYU Abu Dhabi (NYUAD) Center for Interacting Urban Networks, funded by the Abu Dhabi government through the NYUAD Research Institute.

Note on Methodology and Terminology:

This study analyzed a 2025 dataset of 4,283 commercial data centers in the contiguous United States from the commercial Data Center Map database. The dataset includes operational facilities as well as projects that were planned, under construction, or land-banked at the time of analysis. Throughout this article, "city" refers to metropolitan and micropolitan statistical areas (MSAs and MicroSAs), as defined by the U.S. Office of Management and Budget, which classify counties based on economic integration with an urban core rather than population density or land use. As a result, some counties classified as part of metropolitan or micropolitan areas may have low population density or predominantly rural land use. The study characterizes the overall geographic distribution of U.S. data centers by facility count during the study period.

Appendix

Data centers in each metro area - top 10

#	Metro area	Facilities	Share of U.S. total
1	Washington–Arlington–Alexandria	610	14.2%
2	Chicago–Naperville–Elgin	241	5.6%
3	Dallas–Fort Worth–Arlington	192	4.5%
4	New York–Newark–Jersey City	163	3.8%
5	Phoenix–Mesa–Chandler	154	3.6%
6	Atlanta–Sandy Springs–Roswell	136	3.2%
7	Columbus	133	3.1%
8	San Jose–Sunnyvale–Santa Clara	130	3.0%
9	Los Angeles–Long Beach–Anaheim	90	2.1%
10	Des Moines-West Des Moines	76	1.8%

Data centers in Energy Community (EC) designated areas - top 8

#	Metro area	Under-development data centers
1	Chicago–Naperville–Elgin, IL–IN	108
2	Dallas–Fort Worth–Arlington, TX	58
3	Washington–Arlington–Alexandria, DC–VA–MD–WV	26
4	San Antonio–New Braunfels, TX	24
5	Reno, NV	18
6	New Haven, CT	12
7	Scranton–Wilkes-Barre, PA	11
8	Monroe, LA	10

This article was originally published by NYU Tandon School of Engineering and has been republished with permission.

Read next: Study identifies temptation as crucial link in problematic Internet use
by External Contributor via Digital Information World

Study identifies temptation as crucial link in problematic Internet use

By University of Duisburg-Essen. Reviewed by Irfan Ahmad.

Image: Batuhan Doğan - Unsplash

People who turn to online gaming, social media or shopping apps after a stressful day are often looking for distraction or relief. But what determines whether this develops into problematic Internet use? A research team led by Dr. Silke M. Müller at the UDE investigated the everyday online behaviour of 900 adults. Their findings show that while people with problematic Internet use experience higher levels of stress and negative mood, the decisive factor driving them online is temptation. The study* has now been published in PLOS One.

Problematic Internet use refers to a pattern of behaviour in which online activities such as gaming, pornography use, online shopping or social networking become so dominant that they begin to interfere with everyday life. Researchers have long assumed that people turn to the Internet to cope with stress or negative emotions, potentially leading to problematic patterns of use. However, whether these mechanisms actually unfold in everyday life has remained largely unexplored.

Over the past three years, Dr. Silke M. Müller, from the Department of General Psychology: Cognition, and her colleagues from the FOR2974 Research Unit interviewed more than 1,200 adults. Around 900 participants were then followed over a two-week period. Each evening, they completed questionnaires about their mood, stress levels, temptation to go online, time spent online and the consequences of their Internet use. They also reported whether Internet use had brought them pleasure or relief and whether it had caused them to neglect other activities. Before the study began, participants were classified as having non-problematic, risky or pathological Internet use based on standardized diagnostic interviews.

“Our study shows that the temptation to go online is a crucial link between stress, mood and problematic Internet use. It determines whether people engage in online activities – experiencing short-term pleasure or relief while at the same time neglecting other areas of their lives,” says Müller.

Unlike many previous studies, the researchers followed participants in their everyday lives over a period of 14 days. This allowed them to identify why people choose to go online at specific moments. “Understanding these processes is essential for improving prevention and treatment strategies,” explains first author Andreas Oelker. “Our findings suggest that helping people recognize and manage moments of temptation may be more effective than focusing solely on stress or mood.”

* The study is based on data collected within the German Research Foundation (DFG)-funded Research Unit FOR2974, ‘Affective and Cognitive Mechanisms of Specific Internet-use Disorders’. The Research Unit aims to improve understanding of the psychological and neurobiological mechanisms underlying problematic Internet use in order to develop better prevention and treatment strategies. It is led by Prof. Dr. Matthias Brand, who is also a co-author of the PLOS One publication.

Read next: 

• New Report Finds Just 100 Brands Drive a Quarter of All YouTube Sponsorships, and 59% of Brands Have Fewer Than Ten Deals

• Study finds reducing Facebook and Instagram exposure to untrustworthy sources changed feeds but not beliefs

by External Contributor via Digital Information World

New Report Finds Just 100 Brands Drive a Quarter of All YouTube Sponsorships, and 59% of Brands Have Fewer Than Ten Deals

Written by Daniel Moradkhani. Edited by Irfan Ahmad.

An index tracking over 367,994* detected sponsorships reveals a winner-take-most pattern in YouTube brand partnerships.

Most of the leading sponsors are brands familiar to people in the industry, and the average YouTube viewer will recognize them too. These are brands that sponsor creators in a bunch of different niches and have been doing so for the past couple of years consistently. The top 100 brands on YouTube account for 27% of all 367,994 detected sponsorships and the top 10% of brands are behind about 65% of them. The rest of the market is shared by the rest. Basically a winner-take-most market.

GetSponsored is a YouTube sponsorship database built by Stockholm based founder Daniel Moradkhani, tracking 13,944 brands across 25,485 channels. The way it works is it tracks descriptions across thousands of videos, finding out what videos are getting sponsored by what brands. Finding sponsors used to be quite a hassle for Moradkhani. He had to find competitors of his, open each one of their videos in a separate tab, look in their descriptions for links to brands, then do the entire process of finding contacts and pitching himself. So he built GetSponsored, putting it all in one easily manageable place.

"I was opening each video as a tab, looking through the descriptions for brand links, then searching LinkedIn and other sites for contact info for the right individuals," Moradkhani says.

The Brands That Sponsor the Most Creators Are Not the Biggest Companies

NordVPN being on top of the paid sponsorships with 2,150 detected placements across 305 creators won't come as a surprise to anyone.VPNs, specifically NordVPN, have been around for a very long time across tons of niches. Next is Squarespace with 1,831 across 192 creators, then StreamYard (1,441), Shopify (1,417) and finally Hostinger (1,233). BetterHelp, Insta360, Unacademy, Surfshark and Shopee round out the top ten, all somewhere between 1,000 and 1,250 placements. All of these brands are well known in the industry.

The commonality behind these brands is not the company size, instead it's the business model. Almost every single one of them sell some kind of subscription or direct response product where a discount code ties the sale back to a specific video for tracking.

Also read: The Sale That Never Ends in VPNs

Nord Security and Surfshark merged already in 2022 and Incogni is a Surfshark product. Together the three of them sit on roughly 4,000 detected placements. So three different VPN ads on three different channels is often the same parent company three times.

Reach and frequency are also something worth separating. BetterHelp has sponsored 359 different creators which is more than anyone else on the paid side, with 1,226 placements, about three per channel. Whereas Unacademy has almost the same amount of placements, 1,130, spread over just 73 creators, around fifteen per channel. This shows how some brands opt towards niching down when others go for a broader market. Two completely different strategies, one goes wide, one hits the same channels again and again.

Affiliate Links and Paid Campaigns Produce Two Different Leaderboards

Affiliate links are a factor that mildly skew the database. The most notable example of this is Epidemic Sound being the number one brand with 13,393 placements across 298 creators. That's almost as much as the entire paid top 10 combined. Next is Amazon with 8,366 across 352 creators, then Gamersupps (3,692) and Samsung (2,524), and after that a cluster of gaming and hardware brands all between roughly 1,000 and 1,350, Fortnite, Elgato, Epic Games, DJI, Instant Gaming and Apple.

Their reach is basically the same as NordVPN, 298 channels against 305, but they show up about 45 times per channel. NordVPN shows up seven.

"Affiliate links are usually placed across every single one of my videos to maximize revenue. When a brand pays for an integration, the maximal revenue gained is just from placing it in that one video the brand paid for," Moradkhani explains. Which is why brands like Epidemic Sound, who has a very well established affiliate program, lands on top of a lot of sponsorship databases. An affiliate link costs nothing to leave in, so two brands can have more or less the same relationship with the same amount of creators and end up around 11,000 placements apart anyway.

Technology Leads on Volume, VPNs Lead on Intensity

By category Technology is the biggest bucket, 100,662 tracked sponsorships across 1,805 brands. Lifestyle & Vlog comes second at 53,247 across 973 brands, then Entertainment (32,104), Gaming (31,554), Finance & Business (28,757), Beauty & Fashion (25,342) and Health & Fitness (15,260).

Part of this is just that the Technology label is wide. Software, hardware, apps, services, things that would be their own categories anywhere else.

Placements per brand tells a different story. Beauty & Fashion spreads its 25,342 placements across 799 brands, which is around 32 each. VPN & Privacy has 7,706 placements across just 37 brands, more than 200 each. That's the smallest amount of brands of any big category and by far the most active per brand, basically a handful of companies sponsoring at an industrial scale.

Education Apps Are the Fastest Risers While the Largest Paid Sponsors Pull Back

The last 30 days against the 30 before show some movement as well. Two Indian education platforms are the fastest risers, Adda247 grew 146 percent to 113 placements and Unacademy grew 16 percent to 115. In the same amount of time NordVPN appearances dropped about 35%. StreamYard dropped too, around 26%.

These are quite small numbers, low hundreds of placements against an index of 367,994, so this is probably campaign timing more than anything structural. But it does show that not even the heaviest sponsors spend at a constant rate. NordVPN can be number one in the entire index and still be the biggest faller in the same month.

Deal Volume Does Not Track Subscriber Count at the Top of the Index

Of the channels in the index, 823 are under 100K subscribers (3%), 18,311 are between 100K and 1M (72%) and 6,317 are above 1M (25%).

The most sponsored channel in the whole index is John Coogan with 453K subscribers, 796 sponsored videos from 30 different sponsors, more than channels several times his size.

300K viewers that all care about the same thing is a simpler buy than 3M viewers that came for all sorts of reasons. 59% of brands in the index have less than 10 sponsorships.

"From my own inbox it comes in waves. Subscription boxes for a while, then anime apps, then AI chat apps," Moradkhani says.

Methodology and Limitations

The numbers the database represents should be viewed as minimums. Sponsors that might have been removed from the description, exclusively mentioned in the video, or for any other reason not written in the description will not be caught by the detection system. However this is such a minor part of the ecosystem that it practically doesn't matter.

Almost all of the channels in the index are above 100K subscribers, and there are multiple reasons for this. For one, brands usually appear more often in well established channels rather than up and coming ones. "Second, scanning for channels has been done more thoroughly for larger channels. A decent chunk of smaller channels are still currently missing, but will be filled in in the future," Moradkhani says.

Since Epidemic Sound is a brand who has a very well known affiliate program and also does very little real paid advertising in comparison, labeling it as an affiliate brand is mostly correct. However for other brands such as Gamersupps or Samsung, it gets a bit foggier and isn't as accurate yet since the database doesn't verify link by link at this time.

The database started tracking around May 2026, pulling the 15 latest videos of around 74,000 channels, and has been updating daily since then. This potentially leaves a gap for channels who have consistently uploaded for years. Data will most likely deepen in the future.

* Figures are based on GetSponsored's index at the time of reporting and may change as the database continues to update.

About Author: Daniel Moradkhani is the founder of GetSponsored, a YouTube sponsorship database, and runs a network of YouTube channels with more than one billion total views. He is based in Stockholm.

Read next: Study finds reducing Facebook and Instagram exposure to untrustworthy sources changed feeds but not beliefs
by Guest Contributor via Digital Information World