While growing up, at some point, everyone has heard about the Dead Internet Theory, a theory that says most of the internet is controlled by bots. It’s not just you who is scrolling, clicking, and buying things online.

But whenever we think about this, we usually just assume numbers in our head like “60%, 70%, or 80% of internet traffic is bots.”

And with the rise of AI agents, people are talking about this even more.

Until now, these were just assumptions. But now, we actually have some real data — not exactly proving the theory, but showing what bots and AI agents are actually doing on the internet. I was reading a report called The 2026 State of AI Traffic and Cyberthreat Benchmark Report. It is based on the analysis of over one quadrillion digital interactions from across the globe.

So, should we be worried?

My answer is yes. And once you read this article, you’ll understand why.

Let us look at the raw numbers first.

In 2025, automated traffic grew eight times faster than human traffic. While human activity on the web grew by a tiny 3 percent, automated systems surged.

And no, I’m not talking about web crawlers scanning web pages or bots checking news and weather updates. And I think you already know the reason behind this explosion, it’s AI-driven traffic.

Traffic generated by AI systems nearly tripled over the course of the year. It is now the fastest-growing category of internet traffic on earth.

And for the first time ever, AI systems are not just reading the web. They are transacting on it.

To understand what is going on, we need to break this AI traffic down into three distinct categories.

Agents are becoming far more operational. And if this trend continues, these numbers are only going to explode.

You might be wondering who owns all these bots. The answer is heavily concentrated.

OpenAI alone generated approximately 69 percent of all observed AI bot traffic last year. Meta accounted for roughly 16 percent. Anthropic accounted for about 11 percent.

Everyone else combined made up less than 5 percent.

That means the access policies and behavior of just three companies are currently shaping the vast majority of AI traffic on the entire internet.

And this rise of AI agents creates a massive problem for security.

Think about it like this: if I use your credit card, you’d instantly call it fraud. But if an AI agent does the same thing, it’s seen as your personal assistant.

The behavior looks the same. The only real difference is intent.

And according to the data, only about 0.5% separates normal automation from malicious automation.

The old question, “is this a bot or a human?” is dead. Now the real question is: is this action trustworthy or hostile?

And the hostile actors are very, very busy.

Let us look at the four major cyberthreats happening on this new automated internet.

Threat 1: Account Takeovers (ATO)
Account takeover is when a hacker tries to break into your account. They usually buy huge lists of stolen passwords from the dark web and use automated tools to test them across thousands of sites.

Login attacks actually dropped last year, but hackers got much more sophisticated.

Instead of forcing their way in, they’re sneaking in after login. Attacks focused on “post-login compromise” more than quadrupled. This means hackers are abusing session tokens or manipulating account settings to maintain access after a legitimate login happens.

In 2024, the average organization saw around 100,000 of these attempts. In 2025, that number jumped to 402,000.

Threat 2: Carding
Carding is a brute-force financial attack. Hackers use automated tools to run thousands of small transactions to test stolen credit cards. Once a card works, they use it for bigger fraud.

These attacks have surged by 250% compared to 2022.

Now, hackers are using AI browser agents to do this. Researchers saw AI agents run complex sequences, adding a card 11 times, trying payments 6 times, and even switching to redeem loyalty points when cards failed.

Threat 3: Web Scraping
We talked about AI companies scraping data to train models. But malicious scraping is a completely different game.

This is when bad actors or competitors use automation to steal content, bypass paywalls, or mess with pricing at scale.

The numbers are insane, retail alone saw over 150 billion scraping attempts. Globally, nearly 20% of all internet traffic was scraping. And for heavily targeted companies, it goes beyond 60%.

Threat 4: Fake Account Creation
Hackers use automation to create thousands of fake accounts that look legit. They age them, add normal activity, and then use them to drain promo budgets, post fake reviews, or run fraud.

Fake accounts grew 259% in 2024, and another 89% in 2025. One betting platform alone saw 874,000 fake accounts in a year.

Streaming platforms are getting hit too. Bots generate fake streams, likes, and comments, blending in with real users.

Every industry faces a different kind of war on the internet. The report broke down the data by sector, and the results are interesting.

Retail & E-commerce
This is the most targeted sector for scraping. In some cases, over 57% of product page traffic is just scraping. One retailer faced 9.2 billion scraping attempts in a single month. It’s also the top target for carding attacks for the fourth year in a row.

Streaming & Media
For heavily targeted platforms, nearly 71% of login traffic is actually hacking attempts. Scraping is also doubling as AI companies and competitors constantly pull fresh content.

Travel & Hospitality
Carding attacks are rising fast, from just 3% a few years ago to 6× higher today. And weirdly, attacks spike when the weather is nice. One airline saw fraud jump from 1% to 8% of transactions during peak summer.

Financial Services
Anything close to money becomes a prime target. Account takeover attempts are up 39%. In one case, a bank spent 8 months under attack, where 15% of all login traffic was hackers trying to break in.

Technology and SaaS
For software platforms, scraping attacks make up two-thirds of their visits. And for the second year in a row, the outright majority of logins on heavily targeted tech platforms were hackers attempting account takeovers.

The Dark Web Economy
If you want to see how cybersecurity is doing, just look at the dark web. It works like a normal market, pure supply and demand.

When security gets stronger, accounts are harder to steal, supply drops, and prices go up. When security is weak, accounts flood the market, and prices crash.

Right now, tech and finance accounts are expensive. A crypto exchange account can sell for up to $4,500, and email accounts from a major provider have nearly tripled to around $1,000. That means defenses here are getting stronger.

But travel and hospitality tell a different story. One hotel chain’s accounts dropped from $201 to $40, and an airline’s from $175 to $45. That suggests hackers have found an easy, scalable way in.

When you put all these pieces together, a very clear picture emerges.

It’s not just humans using the internet anymore, bots have been part of it for a long time, but with the rise of agentic AI, you can actually feel the shift.

We are moving into an era where software will browse, shop, and manage accounts for us. It will be incredibly convenient. But it will also be very hard to secure.

Organizations can’t simply block all automated traffic. If they do, they risk blocking real AI assistants and losing huge revenue. But if they allow everything, they open the door to fraud, carding attacks, and account takeovers.

In the future, security won’t just be about verifying identity. It will be about verifying intent, as I mentioned earlier.

Systems protecting our data will need to look at an AI agent and decide:
Is this a helpful bot buying shoes for a real person? Or a malicious script testing stolen credit cards?

This shift has already started. We are already living in the agentic internet.

What do you think? Does this change how you see the web, knowing most traffic is software talking to software?

When did you first notice the internet changing?

Remember when the news was about what happened, not how to feel about it? 1440's Daily Digest is bringing that back. Every morning, they sift through 100+ sources to deliver a concise, unbiased briefing — no pundits, no paywalls, no politics. Just the facts, all in five minutes. For free.

Read the newsletter trusted by 4.5 million fact-seekers.