The last real human on the web
In 2025 much of the web is synthetic. How do we verify truth? A guide to provenance (C2PA), proof‑of‑personhood and governance in the AI era.
The AI Assistant Through the Eyes of René Descartes
The internet was once our Library of Alexandria, a direct archive of human knowledge and experience. That archive no longer exists. Instead, we now navigate a hall of mirrors where the reflections no longer originate from reality, but from statistical probabilities. The year 2025 marks a tipping point that Europol analysts feared years ago; an estimated ninety percent of digital content is now synthetically generated or heavily manipulated. We no longer live in an information society, but in an environment analogous to René Descartes’ thought experiment from 1641. His feared ‘Evil Demon’, which presented him with an illusion of reality, is no longer a hypothetical monster. It has become the architecture of the web itself. This fundamental shift in the nature of information has an unprecedented impact on how we perceive and process truth.
The Anatomy of Deception
The urgency of this crisis is not always visible through abstract statistics but sometimes manifests in grotesque, almost laughable incidents that painfully illustrate how deep we have sunk. A striking example of this was the publication in the scientific journal Frontiers in Cell and Developmental Biology. It featured a peer-reviewed article containing an anatomical diagram of a rat. The animal sat upright, possessed four huge testicles, and was surrounded by gibberish labels such as ‘testtomcels’ and ‘dissilced’ [1].
That such an image could pass the heavy gatekeepers of science reveals a terrifying truth. It demonstrates that even experts suffer from a form of epistemic apathy. It looked like science, it had the formatting of knowledge, and so it was accepted as truth. This phenomenon touches directly on the core of modern ethics within technology; when systems become so good at imitating competence, we switch off our critical faculties. It is not that the AI lied, for an AI does not know the concept of truth. The system merely hallucinated a statistically plausible image, and we fell for it.
The Habsburg Internet and Model Autophagy
The pollution of our information ecosystem goes deeper than incidental errors. We are heading towards a phenomenon researchers call ‘Model Collapse’ or, in more vivid terms, the ‘Habsburg AI’. Just as inbreeding in royal families led to genetic degeneration, training new AI models on data from old AI models leads to digital deformities.
When generative models consume primarily synthetic data, nuances, rare facts, and human idiosyncrasies disappear from the output. What remains is a grey, average residue of reality. A study published in Nature already warned that models without fresh, human data are doomed to an irreversible qualitative decline [2]. This creates a vicious cycle where the distinction between fact and fiction not only blurs but where the concept of a ‘fact’ itself is hollowed out by recursive algorithmic repetition. The question “what is true?” is replaced by “what is the most probable average?”.
Math as the New Sense
If our digital senses deceive us and our eyes can no longer distinguish a generated rat from a real one, we must rely on a different set of instruments. In this era of synthetic media, passive perception is no longer sufficient. We must switch to active, cryptographic verification. Here, technical innovation offers the only way out of the hallucination.
The solution lies in standards like the C2PA (Coalition for Content Provenance and Authenticity). Think of this as a digital nutrition label for content. Instead of looking at the image itself, the software looks at the cryptographic signature attached to the file. Was this photo really taken by a Sony camera? Was it edited by Adobe Photoshop? Or was it entirely generated by Midjourney?
This shift means we move the burden of proof. We no longer rely on “I see it, so it is true”, but on the mathematical certainty of the chain of custody. Major technology companies are now starting to integrate these standards deeply into browsers and operating systems. When you see an image in the near future without this digital passport, your instinctive reaction should be one of deep suspicion. The absence of a certificate becomes the clearest signal of potential deception [3].
The Biological Anchoring of the Mind
Besides verifying content, we must also verify the sender. In the digital space, there is no body to prove a mind is present. AI agents can generate the famous statement Cogito, ergo sum (“I think, therefore I am”) a thousand times per second without any awareness of existence.
To prevent the internet from becoming an echo chamber of bots, we are seeing the rise of ‘Proof of Personhood’ systems, such as the controversial but growing World (formerly Worldcoin). Through biometric verification, such as an iris scan, a cryptographic proof is provided that a user is a unique human, without necessarily revealing their identity [4]. This effectively creates a two-tier internet. On one side, the ‘Verified Zone’, where humans talk to humans and where information is traceable. On the other side, the ‘Grey Zone’, a wilderness dominated by bots, hallucinations, and anonymous noise. Although this raises questions about privacy and centralization, it seems to be the only dam against the flood of synthetic noise.
Practical Guide for the Modern Skeptic
How do you navigate this new reality? It requires a fundamental reprogramming of your online behavior. The naive attitude of the early internet years, where information was free and generally reliable, has become dangerous.
- Be lateral, not vertical: Do not read a website from top to bottom, but immediately open multiple tabs to verify the source and the claim via other authorities.
- Distrust perfection: Human communication is messy, erratic, and full of small imperfections. AI texts are often too smooth, too structured, and too rich in buzzwords like ‘pivotal’ and ‘leveraging’. A typo is paradoxically often proof of humanity these days.
- Demand the chain: Do not accept news or images without clear source attribution or cryptographic watermarks. Ask yourself with every post: where is the digital passport?
From Cogito to Verifico
Descartes’ philosophical quest for certainty ended with thinking itself. Our search in 2025 must end with verification. We can no longer afford the luxury of being passive consumers of information. The price for truth in an era of unlimited generation is vigilance.
This requires strict governance on a personal and societal level, where we rely not only on our eyes but on the underlying mathematical proofs. The old adage was Cogito, ergo sum. In the battle against today’s digital demons, a new, harder law applies: Verifico, ergo credo. I verify, therefore I trust. Only by actively reconstructing reality with cryptographic certainty can we escape the hallucination and regain solid ground.
Related signals
References
[1] Oransky I. The “Rat with Four Testicles” Paper and the Crisis of Peer Review. Retraction Watch [Internet]. 2024 [cited 2025 Dec 7]; Available from: https://retractionwatch.com
[2] Shumailov I, Shumaylov Z, Kazhdan Y, et al. The Curse of Recursion: Training on Generated Data Makes Models Forget. Nature [Internet]. 2024 [cited 2025 Dec 7]; Available from: https://www.nature.com/articles/s41586-024-07566-y
[3] C2PA. Technical Specifications for Content Provenance. C2PA.org [Internet]. 2025 [cited 2025 Dec 7]; Available from: https://c2pa.org/specifications/
[4] World. The World Protocol Whitepaper. World.org [Internet]. 2025 [cited 2025 Dec 7]; Available from: https://whitepaper.world.org