The idea that the internet is a void, starved of human contact and orchestrated by AI-generated content and autonomous bots, has never felt more real than in 2026.
Call it the “dead internet theory” or, as Nathan Powell, co-founder and chief product and strategy officer at Fabulate, terms it, “dead creator theory”.
“It challenges the idea of what happens if everyone can make content, but fewer people than ever are trusting what they see,” Powell told Mediaweek at the AiMCO summit.
Nathan Powell Image: Fabulate
Powell says half of web traffic is driven by bots.
“When you look at content itself, more than half of all web-based content has passed through AI algorithms, whether generated, translated, optimised, or otherwise processed,” he said.
The theory, which dates back to 2016 to 2017, is based on the idea that “the internet isn’t as human as we think”.
Hence, the rise of what some describe as the “raw and messy” wave among internet creators and audiences, content that feels less polished, more immediate and more human.
With generative AI, the old notion of “I’ll believe it when I see it” is no longer a reliable test. Seeing is no longer believing.
While much of the AI conversation continues to orbit generative tools, Powell believes the real shift is happening elsewhere.
“We see AI falling into three categories: analytical AI, assistant Gen AI, and agentic AI,” he explained.
The first two, he says, are already embedded in creator marketing.
Analytical AI and brand safety at scale
Analytical AI is increasingly used for brand safety and risk management.
“If you took 10 influencers with an average of 185 videos at 1.5 minutes per video, that’s 46 hours of content to manually review.”
Powell said that AI can scan that back catalogue in minutes.
“Flag instances of brand safety concerns, for example, vaping and timestamp where they occur. There’s always a human in the loop to make the final call, but AI gives you the insurance policy.” Powell said.
Role of Gen AI
Assistant Gen AI, meanwhile, is driving localisation and content augmentation.
Powell pointed to MrBeast as an early example of creator-scale localisation.
“MrBeast famously realised around 35% of his audience was Spanish-speaking and initially hired voice actors to dub his content,” Powell said.
Today, AI can handle translation ethically and at scale, with consent and compensation, allowing creators to extend their reach without re-filming entire productions.
“Brands are producing thousands of content variants weekly by augmenting original creator assets. Image-to-video tools can also generate short B-roll-style clips from still images, saving time without replacing the creative core,” he said.
But Powell argues that the third category, agentic AI, is the real game changer.
“Agentic AI doesn’t just generate content; it acts,” he said.
“If you cut through all the hype of LLMs and generative tools, agentic AI is the real revolution.”
Agentic in practice
In creative marketing, that could mean AI agents filtering hundreds of inbound DMs and brand emails based on predefined parameters.
“It could mean negotiating rates within agreed thresholds and only alerting a creator when human judgment is required. It could mean checking content against compliance guidelines before submission, reducing rounds of revision,” he said.
Powell said the creative remains human; it’s merely the surrounding workflow that becomes automated.
Top Image: Google Gemini