W3C publishes paper on the state of the web in the AI era
The World Wide Web Consortium (W3C) Strategy Team finalised a paper titled ‘W3C and the State of the Web in the AI Era’ on 20 October 2026. The paper was prepared by Philippe Le Hegaret, Vice‑President, Technical Strategy, and will be presented during the Web and AI Inter‑Group session on 29 October 2026 at the Technical Plenary and Advisory Council (TPAC) 2026. The Strategy Team invites feedback from W3C members and the broader community via public‑webai@w3.org and the w3c/webai‑roadmap repository.
The paper notes that, for the first time, more than half of the traffic reaching sites on Cloudflare’s network was not initiated by human interactions with the website. It also states that daily requests from AI agents grew by more than 1 700 % in the year to September 2026. Figures from other vendors are cited: Ruishu Information reports that AI‑agent‑driven traffic rose from below 1 % to between 8 % and 12 % on its network; DataDome records AI accounting for 1.7 % of all requests in June 2026, an increase of 82 % year‑on‑year; and Cloudflare estimates that about 2.5 billion people use generative AI regularly. The paper highlights that AI agents now embed web‑browser functionality and that the web remains the dominant information system for AI.
Risks identified include security (for example, agents accessing private data and potential exfiltration via ordinary text), privacy and consent (cross‑context data use and prompt fatigue), accessibility (possible degradation of ARIA semantics), trust and provenance (synthetic content and impersonation), centralisation and lock‑in, publisher sustainability, equity and bias, and environmental sustainability. Opportunities mentioned include on‑device inference, user‑controlled data via delegated authorisation, accessibility gains such as real‑time captioning, structured‑data use (knowledge graphs, JSON‑LD, RDF, DCAT, ODRL), verifiable credentials and decentralised identifiers, and AI‑assisted threat modelling.
The paper proposes a draft declaration that ‘the web is for all people – including those who choose to use agents’ and that agents acting for a person are user agents that owe loyalty, transparency and safety. It recommends several concrete actions, including:
- chartering an Agentic Web Working Group with an explicit scope for security, privacy, agent identity and delegation;
- protecting ARIA semantics and issuing guidance against agent‑oriented misuse of ARIA;
- capturing AI‑related disclosure, training‑permission and alt‑text needs in the Publishing Maintenance Working Group;
- developing payment requirements for AI agents in the Web Payments Working Group;
- drafting guidance on AI usage and lifecycle impact in the Sustainable Web Interest Group;
- coordinating with external standards bodies such as the IETF, ISO/IEC, C2PA, EMVCo and FIDO.
The paper quotes Bernadette Meehan, chief executive officer of the Wikimedia Foundation, speaking at the United Nations General Assembly in September 2026: ‘the toll of scraping and the behaviour of some of the big technology and frontier AI companies takes a toll on our infrastructure’. No response from the cited technology companies is included in the source.
The development is relevant to digital governance because it seeks to shape web standards that address AI‑driven traffic, security models, privacy consent, accessibility, equity and sustainability, and it notes that some of these areas intersect with emerging regulatory frameworks, such as article 50 of the EU Artificial Intelligence Act. The Strategy Team’s call for feedback and its list of proposed actions illustrate how the W3C intends to engage the technical community in defining standards for the agentic web.
Source: https://www.w3.org/2026/10/w3c-and-state-of-web-in-ai-era.html
