Press kit · Updated August 2026
Facts first. The demo is reproducible.
WARD is an early technical project, not an adopted standard. This page separates what exists now from the thesis being tested, and gives reporters a way to verify the central function themselves.
Boilerplate
Copy-paste description
English
WARD (Web Access Rights Declaration) is an open-source alpha protocol and reference gateway created by Slava Spitsyn in Leipzig, Germany. It turns a public AI access policy into an origin-side allow, observe, or redact decision and can issue a technical receipt describing that decision. WARD explicitly distinguishes a crawler's claimed identity from verified origin evidence. It is experimental, MIT-licensed software—not an official EU, IETF, or W3C standard.
Deutsch
WARD (Web Access Rights Declaration) ist ein quelloffenes Alpha-Protokoll mit Referenz-Gateway, entwickelt von Slava Spitsyn in Leipzig. Eine öffentliche KI-Zugriffsrichtlinie wird am Origin in eine Allow-, Observe- oder Redact-Entscheidung übersetzt und kann als technischer Beleg protokolliert werden. WARD unterscheidet ausdrücklich zwischen behaupteter und am Origin verifizierter Crawler-Identität. Die Software ist experimentell und MIT-lizenziert; sie ist kein offizieller EU-, IETF- oder W3C-Standard.
Verified fact sheet
Current project state
What exists
The reportable technology
Machine-readable policy
A ward.json file declares permissions and per-path overrides.
Open the policy →Explicit crawler uses
Training, search indexing, and user-triggered fetchers are evaluated separately.
Inspect the registry →Origin-side redaction
The reference Next.js site omits marked protected values during server rendering.
Run the canary →Decision receipts
Headers and an optional API describe identity confidence, action, rule, policy hash, and event id.
Read the API status →Important boundaries
Please do not report these as facts
- WARD does not prove that a response was used for model training.
- A User-Agent match does not authenticate the crawler operator.
- WARD does not by itself establish AI Act, copyright, or GDPR compliance.
- WARD has no crawler certification or payment product in the alpha.
- WARD is not currently an IETF, W3C, or EU-recognised standard.
- The public demo does not claim a production publisher rollout.
Why the idea may matter
The thesis being tested
Machine-readable rights reservations are becoming more important, while origin enforcement, requester verification, and audit evidence remain separate technical problems. WARD tests whether a small, interoperable gateway can connect those layers without pretending to solve downstream use or legal interpretation.
EU AI Act, Article 53 ↗
Copyright policy and rights-reservation obligations for GPAI providers.
DSM Directive, Article 4 ↗
Machine-readable reservation of rights for text and data mining.
W3C TDM Reservation Protocol ↗
Existing rights-reservation vocabulary WARD should interoperate with.
IETF AIPREF working group ↗
Emerging preference-expression work; enforcement remains out of scope there.
Assets and interview contact
Updated alpha screenshots, architecture diagrams, founder photo, and a concise backgrounder are available on request. Earlier May materials are archived because the product boundary changed.
info@getward.org