Policy Token
A shared machine-readable signal that allows publishers to embed AI training preferences, which AI companies read and honor when crawling content. It is intended to unify opt-out mechanisms across multiple platforms.
Timeline (last 90 days)
The core mechanism of the new standard is a shared "policy token" signal, into which publishers embed their AI training preferences, and AI companies read and honor it when crawling.
The goal of the standard is to let publishers set their preferences once and have them take effect across multiple AI platforms, without submitting opt-out requests to each company separately.
Compared with robots.txt, the policy token has broader coverage and is designed specifically for AI training scenarios.
The format, transport method and verification mechanism of the policy token require a public technical specification, otherwise publishers cannot actually deploy it.
AI companies need to actually parse and honor the policy token in their crawling pipelines, rather than merely stating that they do so.
In the absence of a violation penalty mechanism, the effectiveness of the new standard depends on industry self-regulation.
No independent third-party audit mechanism has been mentioned so far, which is an uncertain factor in the standard's implementation.
The policy token allows publishers to restrict the use of content for AI training while retaining search indexing, helping to ease the dilemma publishers face between "being seen" and "being used".
The standard currently only involves an opt-out mechanism and does not yet involve a paid licensing framework.
All Facts (9)
The core mechanism of the new standard is a shared "policy token" signal, into which publishers embed their AI training preferences, and AI companies read and honor it when crawling.
50%UnverifiedThe goal of the standard is to let publishers set their preferences once and have them take effect across multiple AI platforms, without submitting opt-out requests to each company separately.
50%UnverifiedCompared with robots.txt, the policy token has broader coverage and is designed specifically for AI training scenarios.
50%UnverifiedThe format, transport method and verification mechanism of the policy token require a public technical specification, otherwise publishers cannot actually deploy it.
50%UnverifiedAI companies need to actually parse and honor the policy token in their crawling pipelines, rather than merely stating that they do so.
50%UnverifiedIn the absence of a violation penalty mechanism, the effectiveness of the new standard depends on industry self-regulation.
50%UnverifiedNo independent third-party audit mechanism has been mentioned so far, which is an uncertain factor in the standard's implementation.
50%UnverifiedThe policy token allows publishers to restrict the use of content for AI training while retaining search indexing, helping to ease the dilemma publishers face between "being seen" and "being used".
50%UnverifiedThe standard currently only involves an opt-out mechanism and does not yet involve a paid licensing framework.
50%