AI is destroying the internet. Math is our only hope.

The internet is no longer a place of objective truth. Recent international conflicts have shown how synthetic content can deceive millions before anyone has a chance to verify the source. We are witnessing a systemic crisis of trust that detectors simply cannot solve. Trying to catch AI with more AI is a losing game because attackers hold the asymmetric advantage in every scenario.

Autonomous AI agents now act on their own. These machines browse the web, execute purchases, and interact with humans without constant oversight. When these agents make errors or engage in fraudulent activities, the damage occurs at a scale that traditional regulation is not built to handle. We have seen medical billing errors and market manipulation that no human operator can fully reconstruct or explain.

Verification is the only path forward. We need a way to prove that an AI system performed specific actions using authorized inputs, without requiring the exposure of private data. Zero-knowledge cryptography provides that solution. By binding outputs to a verifiable process, we can restore integrity to the digital information environment.

This shift moves us from a model based on blind trust to one based on cryptographic certainty. Congress should require that high-risk autonomous systems carry proof of who authorized them and what constraints govern their actions. Liability should attach to the absence of proof, not just the content produced. We secured the early web with HTTPS, and today we must secure the future of the internet with cryptographic proofs.