Search traffic is declining as traditional trust signals break down and sites must prove their credibility from scratch.
AI will not kill websites — it will reboot them. The question is what the near future of the web looks like, where trust can be anchored, and how the chain from AI answer to original source will be assembled.
What is happening to search traffic
Traditional search engines are losing ground. AI summarisation in Google and other search results means users get answers directly on the results page without clicking through to sites. This drives down traffic from search and threatens established content monetisation models.
New infrastructure for AI
The web is adapting to a reality where the primary consumer of content is a model, not a human:
- Crawlers and RAG: next-generation search systems use Retrieval-Augmented Generation to build answers from indexed content
- llms.txt: a standard for machine-readable site descriptions that lets models parse structure and purpose faster
- Agent Readability: content formats optimised for agents rather than people
- MCP (Model Context Protocol): protocols for structured context exchange between models and data sources
- Browser agents: autonomous AI systems that navigate the web independently
The attribution and trust problem
Perplexity and other AI search engines face disputes over sources: models aggregate information, but the chain from answer to original source is often opaque. This creates several challenges:
- How does a user verify an AI answer?
- How do content authors get recognition and traffic?
- How do you distinguish original content from generated?
Markdown layer and the machine web
The idea of a separate layer for models is emerging — a structured Markdown representation of content parallel to human-readable HTML. This would:
- Optimise information transfer to models
- Clearly separate content for people and for AI
- Reduce computational costs of processing
But a risk appears: the web may split into human and machine versions, with different quality and completeness of information.
Verification technologies
Several directions are developing to restore trust:
- C2PA (Coalition for Content Provenance and Authenticity): a standard for cryptographic signing of content to track origin and changes
- Verifiable Credentials: a system of verifiable credentials for authors and sources
- Knowledge Graphs: graphs linking facts to original sources and tracking chains of trust
- AI auditors: specialised models to check the accuracy of answers from other models
What comes next
The web is not dying — it is transforming. Sites become data sources for models rather than the end point of the user journey. Those who succeed will:
- Structure content for machine reading
- Implement verification and attribution standards
- Build a reputation as a reliable source for AI systems
Trust remains central: in a world where AI mediates between user and information, mechanisms for verifying sources must be transparent and reliable.
