To anyone with a pulse and a smartphone, it’s obvious that the internet has an AI slop problem. The issue has grown more severe since ChatGPT launched in 2022, with some social platforms flooded with AI-generated writing. Now, there’s data to back up the anecdotal evidence.

A new preprint study published today from researchers at the Imperial College of London, Stanford University, and the Internet Archive found that approximately 35 percent of all new websites are either AI-generated or AI-assisted. The same study also found that online writing is “increasingly sanitized and artificially cheerful.” In other words, AI is making the internet fake-happy.

The research team tried four different approaches to AI detection before settling on tools from Pangram Labs after it delivered the most consistent results. (Though the team found it performed well on its tests, it is worth noting that all artificial intelligence detection tools are imperfect.) To compile a representative sample of websites, it tapped the Internet Archive’s Wayback Machine, which collects snapshots of webpages. In addition to quantifying how many sites created between 2022 and 2025 lean on AI-generated writing, the study also tested six different theories about the characteristics of slop.

The test that looked into artificial cheerfulness examined how AI affected the tone of online writing. Using sentiment analysis, which classifies words as positive, neutral, or negative, it found that “the average positive sentiment score of AI-generated or AI-assisted was 107 percent higher than that of non-AI websites.” The researchers see this spike in artificial happiness as a “symptom” of the “sycophantic and overoptimistic nature of existing LLMs.” In this way, AI writing tools’ tendency to suck up to their human users has a spillover effect, making the overall tenor of online writing more saccharine.

Another test investigated whether the increase in AI-generated writing shrinks “the range of unique ideas and diverse viewpoints” on offer. The researchers found that AI did make the internet less ideologically diverse, with AI websites scoring roughly 33 percent higher on testing for “semantic similarity” than human-made websites.

While those two tests validated the researchers’ assumptions about AI, others did not. Four theories that the researchers tested were not confirmed. Notably, they had suspected that AI would lead to a rise in misinformation, but their analysis of the evidence did not support the hypothesis. They had also guessed that AI writing wouldn’t link out to external sources, and that it would be stylistically more generic than human writing. Confounding expectations, neither of those theories were supported by the evidence, either.

While the analysis found that the ideas espoused by AI writing were more homogenous—and specifically, more consistently cheery—the writing style itself was not confirmed to be flattened. This came as a big surprise to the researchers, who had assumed they would see a clear move towards more generic output. “Everyone on the team expected that to be true,” says Stanford researcher Maty Bohacek. “But we just don’t have significant evidence for that.”

Prior to conducting its analysis, the research team commissioned a poll on how people feel about AI. Comparing it to the results, it discovered that the researchers weren’t the only ones who had their expectations upended. Many commonly held beliefs about AI writing are wrong, their study finds.

Like the researchers, most people polled had also assumed that they would encounter a rise in fake news as the amount of AI-generated websites they saw increased. The vast majority of respondents had also assumed that AI writing would stop linking to external sources, and that it would have an increasingly generic, uniform voice. “It’s interesting to see that people tended to expect the worst outcomes,” Bohacek says.

This study is far from the last word on what AI is doing to the internet. “We just wanted to break ground,” says Bohacek, who sees this as a jumping-off point for deeper exploration. As a snapshot of AI slop’s impact, it offers a particularly human flavor of insight: Sometimes, it’s simply hard to predict how things will unfold.

Nevertheless, there are clear patterns that appear. In nearly all cases, teenage boys are allegedly responsible for the creation of the images or videos. They are often shared in social media apps or via instant messaging with classmates. And they are hugely harmful to the victims. “I’m worried that every time they see me, they see those photos,” one victim in Iowa said earlier this year. “She’s been crying. She hasn’t been eating,” another’s family said.

In multiple instances, victims often do not want to attend school or be faced with seeing those who created explicit images or videos of them. “She feels hopeless because she knows that these images will likely make it onto the internet and reach pedophiles,” says lawyer Shane Vogt, and three Yale Law School students, Catharine Strong, Tony Sjodin, and Suzanne Castillo, who are representing one unnamed New Jersey teenager in legal action against a nudifying service. “She is severely distressed by the knowledge that these images are out there, and she will have to monitor the internet for the rest of her life to keep them from spreading.”

In South Korea and Australia, schools have given pupils the option not to have their photos in yearbooks or stopped posting images of students on their official social media accounts, citing their use for potential deepfake abuse. “Around the world, there have been cases where school images were taken from public social media pages, altered using AI, and turned into harmful deepfakes,” one school in Australia said. “Imagery will instead feature side profiles, silhouettes, backs of heads, distant group shots, creative filters, or approved stock photography.”

Sexual deepfakes created using AI have existed since around the end of 2017; however, as generative AI systems have emerged and become more powerful, they have led to a shadowy ecosystem of “nudification” or “undress” technologies. Dozens of apps, bots, and websites allow anyone to create sexualized images and videos of others with just a couple of clicks, often with no technical knowledge.

“What AI changes is scale, speed, and accessibility,” says Siddharth Pillai, cofounder and director of the RATI Foundation, a Mumbai-based organization working to prevent violence against women and children. “The technical barrier has dropped significantly, which means more people, including adolescents, can produce more convincing outputs with minimal effort. As with many AI-enabled harms, this results in a glut of content.”

Amanda Goharian, the director of research and insights at child safety group Thorn, says its research indicates that there are different motivations involved in teenagers creating deepfake abuse, ranging from sexual motivations, curiosity, revenge, or even teens daring each other to create the imagery. Studies involving adults who have created deepfake sexual abuse similarly show a host of different reasons why the images may be created. “The goal is not always sexual gratification,” Pillai says. “Increasingly, the intent is humiliation, denigration, and social control.”

“It’s not just about the tech,” says Tanya Horeck, a feminist media studies professor and researcher focusing on gender-based violence who has looked at sexualized deepfakes in UK schools at Anglia Ruskin University. “It’s about the long-standing gender dynamics that facilitate these crimes.”

Taara has announced its light-based wireless optical connectivity technology will power Cintegral’s ST 2110 Fiber-over-Air, enabling real-time TV and media production workflows on remote sets where cable-based infrastructure is unfeasible.

A graduate of X and Google’s Moonshot Factory, Taara has developed technology that uses beams of light to extend high-speed internet to places where traditional infrastructure is difficult to deploy.

Its first system, Taara Lightbridge, is designed to deliver high-speed, secure connectivity across long-range and challenging terrain, helping networks reach farther.

It is claimed to be constructed to deliver bi-directional communication at speeds of up to 20Gbps, and securely transmit data across distances up to 20km while keeping connections “strong and consistent” all the time, “using the energy of a lightbulb” without digging, spectrum licensing or right-of-way permitting. It has already been deployed in more than 20 countries, with operators including Airtel, Digicel, T-Mobile, SoftBank and Liquid.

In addition, Taara’s core Beam technology is designed for operators, enterprises and next-generation data infrastructure, and is attributed with bringing fibre-like speeds to environments where traditional infrastructure is too slow, costly or impractical to build. This is claimed to mark a shift from fixed, physical networks to infrastructure that can evolve at the pace of demand.

In this industry, video footage often has to be stored locally and physically carried to post for transfer, processing and archive. Taara Lightbridge is now being used to create a high-capacity wireless bridge between those locations, allowing production teams to move data in real time across sites without laying an inch of cable.

Cintegral is a production technology specialist working with leading studios and streaming platforms such as Disney, Netflix and Amazon Studios. It has been validating Taara Lightbridge as part of the new ST 2110 fiber over the air offering.

According to Cintegral, Lightbridge enables real-time streaming of high-resolution 4K JPEGXS and 8K RAW video data between on-location and production crews elsewhere on site, helping directors, DOPs, DITs, dailies, editors, VFX, broadcasters and technical teams collaborate during a shoot rather than waiting until each day has wrapped.

“Our goal with ST 2110 Fiber-over-Air is to bring high-performance production workflows to any environment, without being limited by location,” said Cintegral CEO Dane Brehm. “What Taara’s technology enables us to do is extend that capability to places where connectivity would normally be a bottleneck, allowing real-time collaboration between crews, directors and editors on set.”

The collaboration also builds on momentum from the 2026 HPA Tech Retreat, a forum for leaders across media technology, engineering and content creation to explore emerging technologies and trends. At the event, Cintegral showcased Taara Lightbridge and claims to have generated early interest in the use of wireless optical connectivity for advanced production workflows.

For its part, Taara regards the collaboration with Cintegral as marking an important step in its commercial story, showing how wireless optical connectivity can move beyond traditional telecom use cases and into enterprise environments with what it said were “intense demands” for throughput, mobility and real-time collaboration.

Looking at media production, Taara noted that in this use case, teams increasingly need to move large volumes of high-resolution video between locations quickly and reliably, without waiting for fixed-line buildouts or relying on physically transporting storage media.

“You shouldn’t have to dig or lay miles of fiber just to tell a great story,” said Taara founder and CEO Mahesh Krishnaswamy. “With Taara, we aren’t building networks, we’re beaming them. We’re giving production teams the power to deploy fiber-class connectivity out of thin air, exactly when and where the shoot demands it.”