Something is quietly eating the internet. Cloudflare’s public Radar “Bot vs Human” tracker reports that bots account for roughly 57-58 percent of HTTP requests for HTML content, versus about 42-43 percent from humans. Imperva’s Bad Bot Report, using 2025 data, puts bots at about 53 percent of measured web traffic for the second year in a row.

The punchline is not just that more bots exist. It’s that bots are now the dominant “users” of the web’s most basic layer. We are not talking about niche scraping. We are talking about the traffic that powers search, discovery, and ranking. If the majority of requests are automated, then the feedback loop for what gets indexed, recommended, and monetized gets noisier by design. And when that noise is paired with AI writing and AI “answers,” you get a system that can produce volume while eroding meaning.

The source of the anxiety is spelled out by Pangram, an AI detection company, which analyzed websites including LinkedIn, Medium, Twitter, and Reddit. It found that “about one in four long-form items were fully AI-generated.” The details are even more telling: LinkedIn was the most AI-saturated platform, where more than 40 percent of long-form posts were flagged as fully AI-generated. If you broaden beyond “fully AI-generated” to mixed content, X/Twitter was worse off. Pangram reported that almost half of X articles were either fully AI-generated (23.9 percent) or AI-assisted/mixed (22.9 percent), with only 53.2 percent flagged as fully human-authored.

Put those numbers next to the bot traffic split and a second-order problem appears. Platforms already optimize for engagement and freshness. If bots and AI-generated content scale faster than human publishing, ranking systems can become an accelerant, not a filter. You end up with more of what models and systems can easily generate: templated posts, repetitive narratives, and “answers” that sound fluent. The web starts to look less like a marketplace of ideas and more like a production line that can keep running even when it isn’t getting smarter.

This is where the trust issue turns from annoying to operational. The piece argues that AI is not actually intelligent; it is, in practice, “a copy-and-paste scam on an industrial scale,” repeating that it may sound right while often being wrong. It connects that to an important workflow point: the author uses Perplexity not because it is more accurate than other AI LLMs, but because it shows sources. The claim is stark: when you check, the sources are “often crap,” and the risk is that “confidence” makes people accept AI garbage as gospel truth.

There is also a legal and compliance angle in the source that matters for boards. The piece cites a “recent court filing by the New York Times and others” that alleges Vincent Monaco, who leads privacy engineering at OpenAI, acknowledged in a deposition that “OpenAI had searched training datasets and output data despite the company's initial claims that it couldn't access that data.” The outlets also allege OpenAI deleted logs, a violation of court preservation orders. Whether you care about the specific allegation as a matter of legal strategy or as a matter of data governance, the operational takeaway is the same: AI systems are fed by data access patterns and tooling that may conflict with what regulators and customers are told.

For executives, the hardest part is the compounding effect the author describes: “When you pile garbage on top of garbage you do not get reliable information.” The warning is that AI-generated summaries can become the new substrate. Instead of models drawing from primary or reputable secondary sources, the piece says current AI answers increasingly refer to AI summaries, including examples like Google’s AI Overviews. If summaries cite summaries, then errors can replicate at machine speed, and the “ground truth” layer becomes harder to verify. That matters across products, from search and knowledge tools to customer support and research workflows.

The piece also makes clear that there are domains where the trust failure is not just a reputational risk, but a human risk. It says the author would not trust AI answers on serious health problems at all, while recognizing that “millions of people do that every day.” On the company side, that is a liability and brand trust problem. On the product side, it is a measurement problem. On the policy side, it is a governance problem. And on the user side, it is a behavior shift problem: the web is becoming a place where people ask AI for answers and companionship, rather than seeking experts, doing primary research, or engaging with other humans.

So where does that leave decision-makers? You are operating in a world where traffic mix (humans versus bots) and content authenticity (human versus AI-generated) are both changing fast. If bots dominate requests and AI floods feeds, then the “signal” you rely on to drive growth, moderation, ranking, and monetization gets contaminated. Boards should expect more disputes over metrics, more costs in verification and anti-abuse, and more scrutiny over whether AI systems are producing reliable outputs. The strategic stakes are straightforward: if your business depends on trust, discovery, or information integrity, then a web that is increasingly written by AI for AI is not just a content story. It is a measurement, compliance, and product quality story. Pass the cheese, indeed.