The Trust Recession: Why Consumers Are Quietly Opting Out of Believing What They See Online

By Frank Palermo, COO NewRocket

When you cannot trust what you see, what happens? For many people, that is becoming their reality. Whenever they go online, watch videos, look at reviews, like photos, or even read comments from other users on a social media platform, there are seeds of doubt. “Is this even real?” AI has taken those seeds and blown them up. Even a few months ago, it was relatively easy to spot AI generated content. Now, even as someone who works with AI everyday, I can tell you it is even difficult for me to identify whether or not a photo or video has been AI-generated.

Where does that leave us; when we can no longer identify what is real from what has been generated by LLMs, LVMs, or other AI systems? For organizations, it means that consumer trust, which is already extremely hard to build up, and very easy to dismantle, will be even harder to attain. In some ways, it means that word of mouth recommendations will matter more so than ever before. As people continue to seek connections that are grounded in reality, it will mean that the importance of digital content will mean nothing if trust is not the focal point of every organization’s mission.

As information is harder to identify, and as AI images and content continues to disseminate, companies will need to adapt as trust becomes even harder to win.

Looking at the numbers

90% of people reported to be concerned about AI spreading misinformation, according to an August 2024 survey from the Pew Research Center. Of that number, 34% cited that they were extremely concerned. In parallel, in May 2026, 5W identified a 99-point favorability gap between daily AI users (+57) and everyone else (−42), citing the widest behavioral divide in American public opinion.

These numbers are important. They show that for one, more people are worried about AI spreading misinformation than ever before. For another, as some people continue to use AI and have it become a regular part of their day, while others do not, it creates gaps. These are gaps in information, trust, and credibility. These gaps will increase the distrust across the board.

This lack of trust is not limited to politics and societal sentiment. It spans across overall trust in institutions and organizations. In years previous, “seeing is believing” was the status quo online. You could actually get evidence that things did or did not happen when you saw a video, photo, or post online. Now, that is not the case, and organizations need to step up where the gap is forming.

Why AI Is Making False Information Harder to Spot

Part of what makes this moment in time different comes down to accessibility. Previously, producing convincing fake content used to require real skill, time, and resources to stage something believable. GenAI tools have quietly erased those barriers and have made it incredibly easy for anyone to create what they want, in just a few minutes. And as time has gone on, and as AI programs are being constantly trained on new data and trained on how to improve their outputs, the false content is often good enough to pass a casual glance, or even a fairly careful one.

Image: Mirella Callage - Unsplash

Just as important is scale. The output online is not a single, perfectly executed fake slipping through the cracks. Instead we are seeing massive amounts of data that is flooding in all at once. The sheer volume of "good enough" content is making our ability to spot misinformation harder. A fabricated review doesn't need to be flawless if there are a thousand similar ones surrounding it, each reinforcing the others' credibility. One deepfaked spokesperson video doesn't need to survive frame-by-frame analysis if it is shared and reshared without verification.

At the end of the day, volume, not precision, is what erodes trust at scale, because it overwhelms the normal human instinct to verify before believing.

This not only decreases trust, but it decreases tolerance. We will see people just stop going on a website, platform, or engaging in content altogether. Organizations need customers to be engaged in their product or service, so when they go offline, companies will run into serious issues.

What's Actually Driving the Erosion of Trust

As increased exposure of AI-curated content dismantles peoples’ trust and tolerance, it is going to have consequences. One of those is that people will default to suspicion. This will make it even more difficult for organizations to prove their value.

Platform incentives that reward engagement over accuracy will continue to erode trust. As I stated before, trust is very hard to earn and all too easy to destroy in one fell swoop. Organizations and platforms that make it a chore for people to fact-check will be the ones that face the most consequences and loss of business.

Scams, clickbait, and manipulated content, especially at the volume we are seeing, is training us all to be cynical and distrustful of one another and especially of larger institutions and organizations, unless there is a concerted effort to build a bridge of trust before it can crumble. It will be a lot smoother for that bridge to be built now than in the future, because when that trust is eroded, it likely will not be coming back.

How Companies Can Adapt

The starting point is transparency as a baseline; the bridge needs to be built before the trust is gone. You cannot assume an audience will take a claim at face value, brands need to show their work: where information came from, how it was verified, and what process stands behind a given claim. Sourcing and provenance are necessities for organizations to establish that branch of trust.

That transparency, though, only goes so far. Where skepticism is the default, self-attestation carries far less weight than it used to. A company insisting "trust us" is making a claim about itself, not offering evidence. It also shows that you think audiences will accept your claims at face value, which will backfire. Independent validation via third-party audits, outside reviews, verified data, or credible external sources, will shift the burden of proof from "believe our claim" to "here's who checked."

This trust cannot be achieved in one smart campaign. It takes consistency; repeated, verifiable accuracy over time, delivered even when no one is watching closely or asking for it. Only with a proven, dated track record can organizations maintain a level of trust with their customers. But that effort and dedication will pay off when organizations that have invested in maintaining trust continue to thrive, while others lose credibility and business.

User experience and content strategy should start from the premise that the user is skeptical.

This means surfacing verification points naturally, making sourcing visible without requiring extra effort or paywalls to find it, and avoiding design patterns that rely on an audience's willingness to simply believe. In a post-trust internet, the companies that adapt and survive will be the ones that have been maintaining and reinforcing a bridge of trust between themselves and their customers.

Author Bio: Frank Palermo is the Chief Operating Officer of NewRocket, where he helps guide the company’s growth strategy and strengthens its position as a leading advisor in digital workflows, AI, and enterprise transformation. He brings decades of experience building and scaling technology and consulting organizations, with a career that spans software engineering, enterprise platforms, cloud, data, and AI-driven services. Frank is known for combining deep technical fluency with clear operational vision, and for helping clients translate modern technologies into meaningful business outcomes.

Reviewed by by Irfan Ahmad.

Read next:

• 42% of UK Adults Limit AI Use, Mainly Over Privacy and Security Concerns

• Critical thinking has become an AI‑era buzzword. But what does it actually mean, and how do we teach it?
by Guest Contributor via Digital Information World

Critical thinking has become an AI‑era buzzword. But what does it actually mean, and how do we teach it?

Sara Kells, IE University

Image: Keenan Beasley - Unsplash

Spend enough time in discussions about education and artificial intelligence (AI), and critical thinking will inevitably pop up sooner rather than later. AI is changing how students learn and how educators assess learning, and in the face of this shift, critical thinking is often presented as the right pedagogical response.

But as technology reshapes the way we learn and teach, it’s easy to overlook one crucial question: what does critical thinking actually mean in today’s world?

There is a real danger of it becoming a hollow buzzword, an oversimplified, supposedly universal antidote to this new reality in which information, explanations, and increasingly sophisticated outputs are just a click away.

Expanding definitions

Critical thinking is broadly understood as a complex and valuable set of skills and habits. It includes the ability to evaluate evidence, assess arguments, identify assumptions, distinguish stronger claims from weaker ones, and draw reasoned conclusions.

These skills are still vital, but they do not fully capture what students need to confront the cognitive challenges of an AI-powered world.

Recent research has begun exploring this distinction through concepts such as digital critical thinking, which incorporates the idea that online environments, shaped by opaque algorithms, personalisation and platformed information, require people to interpret not only the content they encounter, but also how they ended up seeing it in the first place.

As the environments in which we learn and live evolve, so too should our understanding of critical thinking. As a researcher of civics education, I believe the answers lie not in abandoning traditional definitions of critical thinking, but in expanding them.

Reflection and judgement

Critical thinking happens in two steps. The first is reflection, that vital micro-moment of pause and consideration that comes before the second step of forming a judgement.

Reflection requires people to question evidence, examine assumptions, compare competing interpretations, recognise the limits of their own perspective, and be willing to revise their conclusions in light of stronger arguments or new information.

But modern digital spaces aggressively shape our attention. Digital platforms decide what is visible, trustworthy, and worthy of engagement, then offer content to users in bite-sized videos, short blurbs, and eternally scrollable feeds.

Critical thinking is difficult within these digital spaces because the time and space needed for reflection disappears. Instead, users are likely to skip directly from consumption to judgement.

However, we can still nurture the crucial first stage of reflection outside of these digital environments. And just like any meaningful educational aim, this habit is not acquired through instruction alone. It is cultivated gradually through repeated practice, feedback, reflection, and revision across the curriculum.

While critical thinking begins with disciplined reflection, it does not end there. Reflection prepares us to exercise judgement.

Judgement is where thinking begins to orient action. It is where we decide what deserves our attention, how much confidence to place in our knowledge, and what responsibilities follow from it. It determines how we ultimately participate alongside others in situations where we cannot be 100% certain of what we know.

The key ingredient: intellectual humility

One of education’s less trumpeted achievements is that it helps students discover the limits of their own understanding. Wrestling with difficult ideas, constructing arguments, making mistakes, and revising one’s thinking all do more than just produce knowledge. These processes gradually calibrate judgement by teaching students the difference between reaching an answer and meaningfully understanding a topic.

With AI-powered tools, it is easier than ever to produce work that appears superficially thoughtful, persuasive and sophisticated. A person can sound informed and articulate without ever doing the difficult cognitive work of developing real understanding.

Students must therefore be taught to distinguish between an answer and real understanding. The ability to write fluent, proficiently constructed text is a completely separate skill to crafting a sound argument. Indeed, slick prose can often mask or distract from the absence of clear understanding.

But the risk is not that the use of AI produces students who cannot think. The risk is that it becomes increasingly easy to mistake polished performance for intellectual depth, both in ourselves and in others.

This is where intellectual humility comes in. Neither modesty nor a lack of confidence, this is the ability to recognise the limits of your own understanding, be open to revision, and calibrate confidence based on what you actually know. Intellectual humility is what prevents judgement from hardening into blind certainty.

This ability makes critical thinking more than a collection of cognitive skills. It becomes part of a broader educational project, one that prepares students to exercise judgement responsibly in relation to other people and the shared world they inhabit.

Protecting democracy

Democratic societies depend on people who can weigh competing claims, recognise uncertainty without becoming paralysed by it, and revise their views when necessary.

These capacities do not emerge on their own. But they can develop through educational experiences that repeatedly ask students not only what they think, but also how they reached a particular conclusion, what evidence might lead them to reconsider it, and how much confidence it truly deserves.

Viewed in this broader way, critical thinking cannot be reduced to a singular skill, embedded within a single course, or measured through one assessment. It is cultivated across disciplines, through experiences that force students to revise their conclusions in light of new evidence, defend competing interpretations, explain the reasoning behind their decisions, and reflect honestly on the limits of their own understanding.

In practical terms, this looks like a science experiment that disproves a student’s hypothesis, a history class comparing conflicting interpretations of the same event, or a literature discussion that explores multiple readings of a text.

These kinds of tasks all ask students to practice the same habit: exercising judgement with intellectual humility. The goal is not simply arriving at the correct answer, but learning to recognise when their own thinking should be questioned, refined, or changed.

Ultimately, this habit prepares students for the modern-day real world, where challenges rarely have clear answers or complete information. Like all adults, students will have to navigate competing claims and public debates, often hosted on algorithmically curated platforms filled with AI-generated content. In these spaces, confidence tends to outpace understanding.

Democratic societies need citizens who have practised the difficult work of revising their opinions. The classroom offers a safe, low-stakes environment to train this skill before exercising it in the real world.

Schools and universities have a duty to teach students to exercise sound judgement in the face of uncertainty, disagreement, and complexity. That work begins not with teaching critical thinking as a separate skill, but by designing learning experiences that repeatedly invite students to question, revise, justify, and reconsider their own thinking across every discipline.

By doing this, intellectual humility becomes not just an academic exercise, but a deep-rooted civic habit.

Sara Kells, Director of Program Management at IE Digital Learning and Adjunct Professor of Humanities, IE University

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

Edited by Irfan Ahmad.

Read next: 

• Weak AI regulation may backfire, making products less safe

• Why you let down your guard on ads when scrolling on social media

by External Contributor via Digital Information World

Why you let down your guard on ads when scrolling on social media

Matthew Pittman, University of Tennessee

Image: DaryaDarya LiveJournal - Unsplash

My wife stared at me in shock as I ripped opened the package. The FedEx driver hadn’t even made it back to his truck as I held the item up overhead like a trophy. Fumbling for words, she asked: “Why on earth did you buy a Speedo … with dinosaur tear-away trunks?”

Honestly, I couldn’t remember. Maybe it was for a social media skit? Maybe I was planning on making her laugh? Or maybe I was just a “victim” of an online ad that showed me the wrong message at the right time.

Scroll on social media for even 30 seconds, and your brain is already a little more tired than it was before. And in that mentally foggy state, you become more open to ads that just tell you what to buy – “The facts make it clear, we’re the best!” – than to ads that invite you to draw your own conclusion.

I’m a professor of advertising who has spent years studying how social media affects consumer behavior. And my recent research underscores how the mental tax of processing information, or “cognitive load,” changes the way people respond to product claims from brands.

Explicit versus implicit pitches

Joined by fellow researchers Stan Li and Bixuan Sun, I ran three experiments testing how people processed online ads for eco-friendly products.

In each study, half the participants first spent time on Instagram, scrolling for 30 seconds, while the other half did not. Everyone then viewed an Instagram post from a sustainable brand.

Participants in the first study saw an ad for laundry detergent sheets; the second, an eco-friendly phone case; and the third, a reusable water bottle. Each ad laid out several environmental facts about the product.

The only thing we changed was the last line of the caption. Half the participants saw an implicit conclusion: “Who makes the best, most sustainable detergent? Here are the facts, now decide for yourself.” The other half saw an explicit conclusion: “Who makes the best, most sustainable detergent? The facts make it clear – it’s TruEarth!!”

What emerged was a consistent pattern: Under normal conditions, without cognitive load, participants preferred the implicit conclusion. In other words, they wanted to think for themselves.

But when cognitive load was introduced just through scrolling on Instagram – something half of all U.S. adults do every day – participants preferred the ad that explicitly told them to buy the featured product because it was “the best.”

We found that credibility explained this phenomena. When consumers were under cognitive load and preferred the ads with explicit conclusions, it wasn’t because they consciously noticed that the ad was telling them what to do. Nobody would ever admit to being that impressionable. Rather, they preferred clear and bold conclusions because those claims made the brands seem more credible.

I think this tracks with the way we look at people, too. Under normal conditions, someone who’s bossy or pushy can come across as arrogant and annoying. But when we’re stressed or in a crisis situation, that same assertiveness can be reassuring.

The puzzle is that sustainability claims are challenging for advertisers. You can’t taste “50% lower emissions” or see “ethically sourced” the way you can judge a coffee’s flavor or a shirt’s fit. Sometimes, you have to take the brand’s word for it. That makes credibility especially important. No brand wants to be accused of greenwashing.

The cost of cognitive load

Normally, when people have the mental bandwidth to think something through, they prefer to reach their own conclusions and are less swayed by superficial information. It feels more like their own opinion and less like an order or strong suggestion.

Imagine one friend recommending a restaurant by listing dishes they loved and letting you decide whether it sounds good, while another flatly declares: “You have to go to this restaurant, it’s the best.” We all like to think of ourselves as rational consumers who make up our own minds about what to do, say, and purchase – even if this isn’t true.

And when we have the time, patience and mental energy to really think through a purchase decision, we may prefer ads that let us draw our own conclusion, based on the facts.

But if cognitive load is in play, the calculation flips. Picture trying to evaluate that same restaurant recommendation from your friend while responding to your boss’s texts, half-listening to a podcast and keeping an eye on a pot of pasta on the stove, all at the end of a long week. You just don’t have the spare mental effort to assess the evidence yourself.

In that state, a friend who just tells you to trust them is easier to believe. The confidence sends a signal that they know what they’re talking about, and the extra clarity is a relief rather than an imposition.

That’s essentially what’s happening when people scroll on social media. Their attention is split across a feed of friends, acquaintances, advertisements, family, celebrities, influencers, brands and strangers – all competing for the same limited mental resources. When a sustainability-themed ad follows that scroll, spelling out “it’s us” rather than “you decide,” it doesn’t come across as pushy. It comes across as confident and trustworthy, precisely because users don’t have the spare capacity to work it out on their own.

If you’ve seen the 1991 movie “Father of the Bride,” you’ll know that Steve Martin’s character is a loving father and that Martin Short is not shy. But weddings are stressful. And when Martin’s cognitive load increases to his breaking point, he makes a bad decision:

And this won’t surprise you: Every consumer thinks they are more rational, and less emotional, than everybody else.

Of course, this could be true. Sometimes. But even the most rational person has “foggy brain” now and then. Whether you’re an early bird or a night owl, our research emphasizes the importance of being in the right mental state when buying online, especially for big purchases.

The times when your cognitive load or brain fog is higher – late afternoon, later in the week, or right before a big event – are not when you should be making important financial decisions. And definitely don’t buy anything on a Friday afternoon at 4 p.m.

Instead, you should wait until the following morning, when you can think clearly.

When credibility backfires

While an explicit product claim works because it signals credibility, it can quickly tip into what looks like false advertising when it’s not backed by evidence. Worse, it could trigger a boomerang effect where you end up hating the ad or brand.

In our case, we only tested claims paired with real and specific facts, such as whether the goods used recycled materials and had verified certifications or quantified emissions cuts. But a brand that states its conclusion boldly without backing it up may find that the same trick backfires. And skepticism, once triggered, is hard to undo.

Our finding does have limitations for now. While we studied sustainability messaging in part because it’s uniquely hard to verify on the spot, it’s unclear whether the same effect about explicit messaging shows up for other claims that can be dense and hard to check, such as ads promoting health benefits or offering financial products. We all know how easy it is for false or misleading messages to go viral and do plenty of damage before their claims can be fact-checked.

But for now, if you’re scrolling, half-distracted, and a brand confidently declares itself the best, remember that confidence might be working on you exactly because your guard is down.

So unless you want to experience buyer’s remorse, consider holding off on that late-night purchase for now. Wait until the next morning.

Matthew Pittman, Associate Professor of Advertising and Public Relations, University of Tennessee

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

Reviewed by Irfan Ahmad.

Read next: 

• Research Examines How Visual Appeal, Emotion and Cultural Messages Affect Digital Advertising Outcomes

• Survey Finds 14% of Americans Judge People by Their Phone Background, 33% of Gen Z Do So, 10% Use Default Wallpapers

• Stop policing AI in the classroom, start teaching it

by External Contributor via Digital Information World

Survey Finds 14% of Americans Judge People by Their Phone Background, 33% of Gen Z Do So, 10% Use Default Wallpapers

Reviewed by Irfan Ahmad.

A Talker Research survey released on July 16 found that 14% of U.S. respondents say they secretly judge people based on their phone background. It also found that some see a person's phone background as a reflection of their personality.

The online survey of 2,000 Americans was conducted between June 11 and June 17, 2026. According to the findings, younger respondents were more likely to report judging people by their phone backgrounds. About one-third of Gen Z respondents (33%) said they do so, compared with 17% of millennials.

The survey also looked at the images Americans use as their phone backgrounds. Family members or children topped the list at 19%. Nature or landscape images and personal photos or memories followed at 11% each, while pets and default wallpapers each accounted for 10%.

Among respondents who said they judge phone backgrounds, 32% said a default wallpaper suggests someone is practical and minimalist, while 25% said it shows a phone is used mainly as a tool. Another 22% said a default wallpaper does not mean anything, while 12% said it suggests the person is uninspired or lacks creativity.

Chart: Phone Backgrounds Used by Survey Respondents. Credit: DIW. CC BY

Read next: Stop policing AI in the classroom, start teaching it
by AI Analysis via Digital Information World

Stop policing AI in the classroom, start teaching it

Zakaria Lacheheb, International Islamic University Malaysia

When handheld calculators entered classrooms in the 1970s, critics sounded the alarm on a wave of math illiteracy. With the internet, the panic shifted to the death of reading.

Yet, neither apocalypse happened. Instead, education adapted, turning these disruptors into essential extensions of human intelligence.

Generative AI marks a similar turning point for higher education. In lecture halls across Malaysia and Indonesia, students routinely use tools like ChatGPT, Gemini, and Claude for everything from research and writing to coding and problem-solving .

Many universities are pushing back against this trend, introducing stricter regulations, academic penalties, and AI-detection software.

This response fundamentally misjudges the challenge. The question isn’t whether students should use AI — they already do. The real question is how universities will train them to use it responsibly.

Instead of fighting generative AI as a threat to integrity, institutions in Malaysia and Indonesia should embrace it as a vital productivity tool by focusing on three priorities: teaching AI literacy, redesigning assessments, and embedding ethical frameworks.

AI as a productivity skill

Future graduates will enter workplaces where AI assists with predictive analytics, software engineering, financial analysis, legal research, and strategic decision-making. More than one-third of entry-level jobs now require AI skills, nearly triple the share from fall 2025. Some 28% of employers say they’re seeking early-career talent who can use AI in their work.

Image: Kojo Kwarteng - Unsplash

Universities should therefore teach AI as a professional competency. Students should learn how to construct effective prompts, evaluate machine-generated outputs, detect hallucinations, verify evidence, and recognise the limitations of AI systems. These are no longer optional digital skills but essential competencies for an AI-enabled workforce.

This approach is consistent with UNESCO’s call for a human-centred model of AI governance that prioritises AI literacy over blanket prohibition. Rather than discouraging students from using AI, universities should equip them to use it critically and responsibly.

Malaysia and Indonesia have an opportunity to lead this transformation. Indonesia’s recent Joint Ministerial Decree signed by seven ministers represents an important step in regulating AI across education. While the policy places greater restrictions on the use of instant-answer AI in primary and secondary schools, it also highlights the need for universities to develop clearer operational frameworks for responsible AI adoption.

Malaysia, with its ambitions to become a regional digital and knowledge economy, should similarly position higher education as a driver of AI literacy rather than AI avoidance.

Universities should encourage students to move beyond using AI merely to generate assignments. They should become creators of AI-driven solutions.

In Malaysia, for example, Islamic economics and finance students could develop specialised AI applications trained on key concepts like Maqasid al-Sharia (the framework for human wellbeing and justice), waqf (charitable endowments), zakat (obligatory almsgiving), and Sharia-compliant investment principles.

AI should become a platform for innovation rather than simply a shortcut for completing coursework.

Redesign assessment for the AI era

When assignments merely reward memorisation, generative AI feels like a threat. The real challenge isn’t the technology, it’s how we design assessments.

Universities should shift away from traditional take-home essays towards assessments that measure reasoning, judgement, and application. Greater emphasis should be placed on oral presentations, live case analyses, project demonstrations, reflective portfolios, and collaborative innovation challenges.

Where AI is used, students should be assessed on how they engaged with it. What prompts did they design? How did they identify inaccuracies or misinterpretation? Which sources did they use to verify AI-generated information? How did AI contribute to, rather than replace, their own analysis?

By assessing the learning process instead of merely the final written product, universities can preserve academic integrity while encouraging responsible technological adoption.

Embed ethical frameworks

Teaching AI without ethics leaves graduates unprepared for the real-world responsibilities of these technologies. This is why universities may need to twin AI literacy with ethical literacy.

For Malaysia, this dialogue can be grounded in the principles of Maqasid al-Sharia, which hold that any policy or practice must be evaluated by its ability to protect five core human goods: faith, life, intellect, lineage, and wealth.

AI tools that drive financial inclusion, alleviate poverty, or simplify complex policy preserve intellect and wealth. On the other hand, using AI to fake research, deceive lecturers, or bypass critical thinking directly violates these principles.

Ethical AI education should not be limited to preventing misconduct. It should cultivate graduates who understand transparency, accountability, intellectual honesty, and the social responsibilities that accompany technological innovation.

Ultimately, universities should not aspire to produce graduates who simply know how to use AI. They should produce graduates who know when to trust it, when to question it, and when human judgement must prevail.

Preparing graduates, not policing technology

The future of higher education in Malaysia and Indonesia should not be defined by increasingly sophisticated methods of detecting AI use. It should be defined by increasingly effective ways of teaching students to use AI wisely.

Teaching AI as a productivity skill, redesigning assessments to reward critical thinking, and embedding ethical responsibility will better prepare graduates for an economy in which AI is becoming an everyday professional tool.

The universities that lead the next generation of higher education will not be those that police AI the hardest. They will be the ones that empower students to use it more intelligently, creatively, and ethically.

Zakaria Lacheheb, Assistant Professor, International Islamic University Malaysia

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

Reviewed by Irfan Ahmad.

Read next: 

• US and Chinese companies train almost all of the world’s most-used AI models

• Upwork Reports Growth in Skilled Freelancing as AI Reshapes the Value of Work

by External Contributor via Digital Information World

US and Chinese companies train almost all of the world’s most-used AI models

By Edouard Mathieu, Head of Data and Research at Our World In Data

Dozens of companies worldwide develop large AI models, but it can be difficult to get a sense of where the most-used ones tend to come from.

OpenRouter is a large platform that allows users to interact with and write software on top of AI models through a single interface. It includes all the models from large companies like Google, OpenAI, Anthropic, Meta, DeepSeek, and Alibaba, as well as many, many others.

I analyzed the data published by OpenRouter on the 50 most used models each day since January 2025 and calculated the average monthly presence by origin country of the models.

As you can see on the chart, US-based companies still account for most models in OpenRouter’s top 50. But their presence has declined, and China-based companies have grown rapidly, from 5 models in the daily top 50 at the beginning of 2025 to 20 in May 2026.

Very few top-50 models come from companies outside the United States and China. Canada was represented early in 2025 by Cohere’s Command R models, while France remains represented by Mistral AI’s NeMo model.

A technology that more people use every year is, so far, almost entirely the product of two countries.

Originally published by Edouard Mathieu at Our World in Data. Republished here under a Creative Commons license.

Reviewed by Irfan Ahmad.

Read next: 

• Upwork Reports Growth in Skilled Freelancing as AI Reshapes the Value of Work

by External Contributor via Digital Information World

Four ways to help your teen (and yourself) spend more time away from devices

Danielle Einstein, Macquarie University

Image: Hannah Busing - Unsplash

Phones and devices have become inextricably linked with everyday life. They store our credit cards, provide critical bus updates, and allow us to communicate whenever we need to.

But when our device use starts to affect our mood, replace real-life experiences, or interfere with face-to-face relationships, it’s a sign our habits have crossed the line from healthy to unhealthy.

How are teens more affected?

Teenagers are particularly vulnerable to excessive device use, as the part of the brain responsible for planning, imagining consequences and working for delayed rewards has not yet fully matured. They also have heightened sensitivities to rewards and social evaluation (in other words, what your peers think of you).

Research shows teenagers are more prone to mood swings, are still evolving their ability to manage uncertainty and are learning how to regulate their emotions.

All of this makes them more susceptible to online triggers that can trap a vulnerable teen in a cycle of conflicting and constantly changing emotions, particularly when information is arriving constantly.

For teens (and adults) it can become a complete world of its own where daily responsibilities are avoided and challenges are only faced with a dependency on phones.

So how do you know when it’s become a problem?

What to look out for

There are some obvious signs excessive device use is becoming a problem for teens (or yourself).

1. They use phones to fill in spare time. If every quiet moment is filled with scrolling, checking notifications or opening apps, it shows a teen’s device is the primary way they respond to spare time, stress or discomfort.

2. They appear distracted when you talk to them. If your teen automatically reaches for their phone midway through conversation, it might be a sign they devalue face-to-face exchanges and have lost the discipline to wait.

3. They put off important tasks. When teens routinely avoid everyday responsibilities such as work, homework, school, household chores or tasks that involve mixing with new people.

4. They seem tired all the time. Teens may block negative thoughts by scrolling in bed, socialising online late in the night, or even getting into bed after the school day.

How does it get to this stage?

The devices we rely on have the potential to develop what I call the “addictive pull” due to the features and services available on them. The “pull” begins when notifications, such as likes, messages, loot box wins and emails, arrive on devices unpredictably. Some are positive while others are neutral. And some also relieve worries. This creates a powerful dopamine-driven loop.

Over time, the device itself becomes a conditioned stimulus. This means even something as simple as seeing your phone case, or watching your device’s screen light up, is enough to trigger an urge to check it. It’s almost as if we are magnetically pulled to the device in the hope of a reward (such as more messages or likes), or to soothe a preoccupying worry.

This can also create an anxiety loop which means even without a notification, a person checks their device, rechecks it and experiences brief moments of relief via a message or app. When the loop occurs on repeat, it winds up worry and creates emotional dependence on others or apps.

The “addictive pull” can be hard to resist. Over time these unnoticed habits can become associated with the space where devices are often used (such as the bedroom, the apartment, the bus). The “pull” to re-engage with a device is a result of predictable conditioning and reinforcement processes rather than a lack of willpower. The bedroom – a place of rest, privacy and study – can be particularly problematic for teens and adults.

Leading by example

We cannot expect children or teens to break from the “addictive pull” if the adults around them don’t either.

As a parent, these are some ways you can lead by example to be more intentional when using devices.

1. Recognise the subtle tension that builds when you have a worry and want to reach for your device immediately.

2. When you walk into your home, place your phone or smartwatch out of the way and in a bag.

3. Do not have your phone or tablet in arm’s reach when it is designated family time.

4. Work with each family member to put apps being used without restraint onto one device per person. For instance, a parent or older teen might have TikTok on a tablet and not on their phone (so they can use their phone without distraction). They should then try to use that tablet in only one room of the house, outside of the bedroom.

Remember, an honest picture of everyone’s screen time habits sit on the device’s screen time records. We may get a minor fright when looking at it, but rather than resigning ourselves to this new way of life – and ignoring the insidious impact on attention, mood and wellbeing – we can commit to one another to make small changes.

Danielle Einstein, Adjunct Fellow, School of Psychological Sciences, Macquarie University

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

Reviewed by Irfan Ahmad.

Read next: 

• Young people are among the most at risk of stalking – but many don’t recognise it

• Research Examines How Visual Appeal, Emotion and Cultural Messages Affect Digital Advertising Outcomes

by External Contributor via Digital Information World