<?xml version='1.0' encoding='utf-8'?>
<!DOCTYPE rfc [
  <!ENTITY nbsp    "&#160;">
  <!ENTITY zwsp   "&#8203;">
  <!ENTITY nbhy   "&#8209;">
  <!ENTITY wj     "&#8288;">
]>
<?xml-stylesheet type="text/xsl" href="rfc2629.xslt" ?>
<!-- generated by https://github.com/cabo/kramdown-rfc version 1.7.43 (Ruby 3.4.12) -->
<rfc xmlns:xi="http://www.w3.org/2001/XInclude" ipr="trust200902" docName="draft-kelley-aipref-training-preference-signal-01" category="std" consensus="true" submissionType="IETF" tocInclude="true" sortRefs="true" symRefs="true" version="3">
  <!-- xml2rfc v2v3 conversion 3.35.0 -->
  <front>
    <title abbrev="Training Preference Signal">Machine-Readable Training Preference Signal</title>
    <seriesInfo name="Internet-Draft" value="draft-kelley-aipref-training-preference-signal-01"/>
    <author fullname="Kevin Kelley">
      <organization>Anthropic</organization>
      <address>
        <email>kelleyk-ietf@kelleyk.net</email>
      </address>
    </author>
    <date year="2026" month="November" day="16"/>
    <area>Web and Internet Transport</area>
    <workgroup>AI Preferences</workgroup>
    <keyword>robots.txt</keyword>
    <keyword>AI preferences</keyword>
    <keyword>AI training</keyword>
    <abstract>
      <?line 40?>

<t>This document describes how a service operator binds usage preferences
(e.g., "no AI training") to paths by writing ordinary <tt>robots.txt</tt>
groups keyed by a per-purpose policy token in the <tt>User-Agent</tt> line.
It introduces no new directive names and no new grammar: the path
language, longest-match precedence, and group structure of <tt>robots.txt</tt>
are not changed.</t>
    </abstract>
    <note removeInRFC="true">
      <name>About This Document</name>
      <t>
        Status information for this document may be found at <eref target="https://datatracker.ietf.org/doc/draft-kelley-aipref-training-preference-signal/"/>.
      </t>
      <t>
        Discussion of this document takes place on the
        AI Preferences Working Group mailing list (<eref target="mailto:ai-control@ietf.org"/>),
        which is archived at <eref target="https://mailarchive.ietf.org/arch/browse/ai-control/"/>.
        Subscribe at <eref target="https://www.ietf.org/mailman/listinfo/ai-control/"/>.
      </t>
    </note>
  </front>
  <middle>
    <?line 50?>

<section anchor="intro">
      <name>Introduction</name>
      <t>A service operator binds usage preferences (e.g., "no AI training") to
paths by writing ordinary <tt>robots.txt</tt> groups keyed by a
per-purpose <strong>policy token</strong> in the <tt>User-Agent</tt> line.</t>
      <figure>
        <name>A policy-token group</name>
        <artwork><![CDATA[
User-Agent: __usage_ai-training
Disallow: /
Allow: /press-kit/
]]></artwork>
      </figure>
      <t>This expresses a preference <em>against</em> AI training for all paths
except those under <tt>/press-kit/</tt>, for which it expresses a preference
<em>in favor</em> (<xref target="checking"/>).</t>
      <t>This document introduces no new directive names and no new grammar; the
path language, longest-match precedence, and group structure defined by
<xref target="ROBOTS"/> are not changed.  As a result, the wide variety of existing
parsers deployed across the Internet will continue to interpret
traffic management signals as they did previously, and will ignore
policy-token groups as they would any group for a user agent they do
not recognize.</t>
      <section anchor="goals">
        <name>Goals</name>
        <ol spacing="normal" type="1"><li>
            <t><strong>Path-specific bindings.</strong> Preferences can be expressed per path,
and not only site-wide (<xref target="checking"/>).</t>
          </li>
          <li>
            <t><strong>Minimal distance from RFC 9309.</strong> Semantics, and ideally
syntax, stay as close to <xref target="ROBOTS"/> as possible, so that authors and
implementers can use the mental model they already have.</t>
          </li>
          <li>
            <t><strong>No large-scale duplication.</strong> <tt>robots.txt</tt> files are
size-constrained (<xref section="2.5" sectionFormat="of" target="ROBOTS"/>), and some deployed
files already contain heavily duplicated groups.  Expressing
preferences does not require restating existing rules
(<xref target="policy-token"/>).</t>
          </li>
          <li>
            <t><strong>Unchanged legacy behavior.</strong> A parser or consumer that does not
implement this mechanism behaves as before.  This holds for parsers
that are actually deployed, not only for idealized RFC 9309
parsers.  Conversely, implementations treat files that do not use
this mechanism exactly as before (<xref target="checking"/> and
<xref target="independence"/>).</t>
          </li>
          <li>
            <t><strong>Consistent interpretation across consumers.</strong> This mechanism does
not depend on the divergent parts of deployed parser behavior
(<xref target="policy-token"/>).</t>
          </li>
        </ol>
      </section>
    </section>
    <section anchor="conventions-and-definitions">
      <name>Conventions and Definitions</name>
      <t>The key words "<bcp14>MUST</bcp14>", "<bcp14>MUST NOT</bcp14>", "<bcp14>REQUIRED</bcp14>", "<bcp14>SHALL</bcp14>", "<bcp14>SHALL
NOT</bcp14>", "<bcp14>SHOULD</bcp14>", "<bcp14>SHOULD NOT</bcp14>", "<bcp14>RECOMMENDED</bcp14>", "<bcp14>NOT RECOMMENDED</bcp14>",
"<bcp14>MAY</bcp14>", and "<bcp14>OPTIONAL</bcp14>" in this document are to be interpreted as
described in BCP 14 <xref target="RFC2119"/> <xref target="RFC8174"/> when, and only when, they
appear in all capitals, as shown here.</t>
      <?line -18?>

<dl>
        <dt>service operator:</dt>
        <dd>
          <t>An entity that operates an HTTP service, such as a website, and
publishes a <tt>robots.txt</tt> file for it.  For preferences expressed
through this mechanism, the service operator is the "declaring
party" (<xref section="2" sectionFormat="of" target="VOCAB"/>).  <xref target="ROBOTS"/> calls this party a
"service owner".</t>
        </dd>
        <dt>consumer:</dt>
        <dd>
          <t>An entity that interprets preference signals that a service operator
expresses through the mechanism defined in this document.</t>
        </dd>
        <dt>parser:</dt>
        <dd>
          <t>In the sense of <xref target="ROBOTS"/>; software that parses <tt>robots.txt</tt> files
and matches product tokens and paths against their rules.</t>
        </dd>
        <dt>policy token:</dt>
        <dd>
          <t>A product token (<xref section="2.2.1" sectionFormat="of" target="ROBOTS"/>) whose RFC 9309 signals
a consumer uses to determine how content will be used.  A policy
token need not appear in the <tt>User-Agent</tt> header field
(<xref section="10.1.5" sectionFormat="of" target="HTTP"/>) of any outbound request.  This document
introduces a new, cross-vendor policy token (<xref target="policy-token"/>).</t>
        </dd>
        <dt>proprietary policy token:</dt>
        <dd>
          <t>A policy token documented by a single consumer.  Some service
operators already include these in their <tt>robots.txt</tt> files.</t>
        </dd>
      </dl>
    </section>
    <section anchor="vocabulary">
      <name>Vocabulary</name>
      <t>The <tt>__usage_ai-training</tt> policy token corresponds to the definition
that the IETF aipref Working Group has arrived at
(<xref section="4.1" sectionFormat="of" target="VOCAB"/>).  The following definitions are quoted from
that document:</t>
      <dl>
        <dt>AI Training:</dt>
        <dd>
          <t>Using an asset to modify the learned parameters of a generative AI
model.
</t>
          <t>The training of models that are used to perform exclusively
non-generative tasks is not included in this category, even if the
model is capable of generative tasks.</t>
        </dd>
        <dt>Generative AI model:</dt>
        <dd>
          <t>An AI model that is used to generate synthetic content in one or more
modalities. Alternatively, an AI model that is made available for use
to generate synthetic content in one or more modalities. This
definition does not include a model that is used, or made available
for use, for classification, ranking, or scoring, even where the model
generates rationale for its output.</t>
        </dd>
      </dl>
      <t>This document applies those definitions as follows:</t>
      <dl>
        <dt>asset:</dt>
        <dd>
          <t>The content retrieved from a path.  The preferences that apply to an
asset are those expressed in the <tt>robots.txt</tt> file in effect when the
asset was retrieved (<xref section="2.4" sectionFormat="of" target="ROBOTS"/>; see also <xref target="ATTACH"/>).</t>
        </dd>
      </dl>
      <t>This document does not define a parallel vocabulary; it tracks the
work of the aipref Working Group.  Implementers <bcp14>SHOULD</bcp14> consider
updating their definitions, and adding or removing vocabulary terms,
as the aipref Working Group reaches consensus in additional areas.</t>
    </section>
    <section anchor="design">
      <name>Design</name>
      <section anchor="policy-token">
        <name>The Policy Token</name>
        <t>This document defines the policy token <tt>__usage_ai-training</tt>.</t>
        <t>The policy token conforms to the product-token syntax in
<xref section="2.2.1" sectionFormat="of" target="ROBOTS"/> (letters, <tt>_</tt>, <tt>-</tt>).</t>
        <t>Consumers <bcp14>MUST</bcp14> match policy tokens exactly, apart from case, as for
any product token.  (Note that some deployed parsers match more
flexibly; the <tt>__</tt> prefix makes accidental matches unlikely.)</t>
        <t>A single group <bcp14>MAY</bcp14> name several policy tokens (consecutive
<tt>User-Agent</tt> lines; <xref section="2.2" sectionFormat="of" target="ROBOTS"/>) to bind its rules to
several purposes.  Policy tokens and other product tokens <bcp14>MAY</bcp14> be
mixed in a group.</t>
      </section>
      <section anchor="checking">
        <name>Checking for Preferences</name>
        <t>To determine whether a preference has been expressed for a given path,
a consumer looks up the path and policy-token combination as specified
in Sections <xref target="ROBOTS" section="2.2.1" sectionFormat="bare"/> and <xref target="ROBOTS" section="2.2.2" sectionFormat="bare"/> of <xref target="ROBOTS"/>.</t>
        <t>However, wildcard groups (i.e., <tt>User-Agent: *</tt>) <bcp14>MUST NOT</bcp14> be interpreted
as matching the policy token.</t>
        <t>If the lookup finds that the path is disallowed, a preference against
the use described by the policy token has been expressed for that path.
If it finds that the path is allowed, a preference in favor of that use
has been expressed.</t>
        <t>If no group contains the <tt>__usage_ai-training</tt> policy token (i.e.,
there is no <tt>User-Agent: __usage_ai-training</tt> line in the file), then
no preference has been expressed.  If such a file contains a
proprietary policy token, consumers <bcp14>SHOULD</bcp14> interpret that token as
they would in the absence of this document, but <bcp14>MUST NOT</bcp14> treat it as
expressing a preference in the context of this document (see
<xref target="precedence"/>).</t>
      </section>
      <section anchor="independence">
        <name>Independence from Crawler Rules</name>
        <t>Consumers <bcp14>MUST</bcp14> interpret the RFC 9309 signals being expressed based on
their own product token(s) exactly as they would in the absence of
this document.  That is, <tt>ExampleBot</tt> still checks whether there are
rules for <tt>ExampleBot</tt>; if there are, they still express the service
operator's preference about crawling.  Nothing in this document
changes that signaling.</t>
        <t>Content-usage preferences and traffic management signals can both be
expressed in <tt>robots.txt</tt>, but they are independent and orthogonal.</t>
      </section>
      <section anchor="crawling">
        <name>Effects on Crawling Behavior</name>
        <t>This policy token is not a directive for crawling behavior.</t>
        <t>A consumer <bcp14>MAY</bcp14> use preferences to inform crawling.  For example, if it
crawls only for purposes covered by the policy token, it <bcp14>MAY</bcp14> refrain
from retrieving resources for which a preference against those
purposes has been expressed.</t>
        <t>If a service operator wishes to limit crawling, it <bcp14>MAY</bcp14> use groups for
crawlers' product tokens in <tt>robots.txt</tt> to do so.</t>
      </section>
      <section anchor="precedence">
        <name>Precedence</name>
        <t>Specific expressions <bcp14>SHOULD</bcp14> take precedence over general ones.</t>
        <t>Consumers <bcp14>MAY</bcp14> document that they interpret rules that match
proprietary policy tokens as being more specific than the
<tt>__usage_ai-training</tt> policy token.</t>
        <t>Private agreements <bcp14>MAY</bcp14> override preferences expressed via this
mechanism.</t>
      </section>
    </section>
    <section anchor="security">
      <name>Security Considerations</name>
      <t>The security considerations of <xref section="3" sectionFormat="of" target="ROBOTS"/> apply to this
mechanism.</t>
    </section>
    <section anchor="privacy">
      <name>Privacy Considerations</name>
      <t>This document does not introduce any new privacy considerations.</t>
    </section>
    <section anchor="iana">
      <name>IANA Considerations</name>
      <t>This document has no IANA actions.</t>
    </section>
  </middle>
  <back>
    <references anchor="sec-combined-references">
      <name>References</name>
      <references anchor="sec-normative-references">
        <name>Normative References</name>
        <reference anchor="ROBOTS">
          <front>
            <title>Robots Exclusion Protocol</title>
            <author fullname="M. Koster" initials="M." surname="Koster"/>
            <author fullname="G. Illyes" initials="G." surname="Illyes"/>
            <author fullname="H. Zeller" initials="H." surname="Zeller"/>
            <author fullname="L. Sassman" initials="L." surname="Sassman"/>
            <date month="September" year="2022"/>
            <abstract>
              <t>This document specifies and extends the "Robots Exclusion Protocol" method originally defined by Martijn Koster in 1994 for service owners to control how content served by their services may be accessed, if at all, by automatic clients known as crawlers. Specifically, it adds definition language for the protocol, instructions for handling errors, and instructions for caching.</t>
            </abstract>
          </front>
          <seriesInfo name="RFC" value="9309"/>
          <seriesInfo name="DOI" value="10.17487/RFC9309"/>
        </reference>
        <reference anchor="VOCAB">
          <front>
            <title>A Vocabulary For Expressing AI Usage Preferences</title>
            <author fullname="Paul Keller" initials="P." surname="Keller">
              <organization>Open Future</organization>
            </author>
            <author fullname="Martin Thomson" initials="M." surname="Thomson">
              <organization>Mozilla</organization>
            </author>
            <date day="13" month="September" year="2026"/>
            <abstract>
              <t>   This document defines a vocabulary for expressing preferences
   regarding how digital assets are used by automated processing
   systems.  This vocabulary allows for the declaration of restrictions
   or permissions for use of digital assets by such systems.

              </t>
            </abstract>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-ietf-aipref-vocab-08"/>
        </reference>
        <reference anchor="RFC2119">
          <front>
            <title>Key words for use in RFCs to Indicate Requirement Levels</title>
            <author fullname="S. Bradner" initials="S." surname="Bradner"/>
            <date month="March" year="1997"/>
            <abstract>
              <t>In many standards track documents several words are used to signify the requirements in the specification. These words are often capitalized. This document defines these words as they should be interpreted in IETF documents. This document specifies an Internet Best Current Practices for the Internet Community, and requests discussion and suggestions for improvements.</t>
            </abstract>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="2119"/>
          <seriesInfo name="DOI" value="10.17487/RFC2119"/>
        </reference>
        <reference anchor="RFC8174">
          <front>
            <title>Ambiguity of Uppercase vs Lowercase in RFC 2119 Key Words</title>
            <author fullname="B. Leiba" initials="B." surname="Leiba"/>
            <date month="May" year="2017"/>
            <abstract>
              <t>RFC 2119 specifies common key words that may be used in protocol specifications. This document aims to reduce the ambiguity by clarifying that only UPPERCASE usage of the key words have the defined special meanings.</t>
            </abstract>
          </front>
          <seriesInfo name="BCP" value="14"/>
          <seriesInfo name="RFC" value="8174"/>
          <seriesInfo name="DOI" value="10.17487/RFC8174"/>
        </reference>
      </references>
      <references anchor="sec-informative-references">
        <name>Informative References</name>
        <reference anchor="ATTACH">
          <front>
            <title>Associating AI Usage Preferences with Content in HTTP</title>
            <author fullname="Gary Illyes" initials="G." surname="Illyes">
              <organization>Google</organization>
            </author>
            <author fullname="Martin Thomson" initials="M." surname="Thomson">
              <organization>Mozilla</organization>
            </author>
            <date day="18" month="August" year="2026"/>
            <abstract>
              <t>   Methods are defined for associating usage preferences with content
   that is obtained using the HTTP protocol.  This document defines
   attachment methods using the Robots Exclusion Protocol and HTTP
   header fields.

   This document updates RFC 9309 to allow for the inclusion of usage
   preferences.

              </t>
            </abstract>
          </front>
          <seriesInfo name="Internet-Draft" value="draft-ietf-aipref-attach-05"/>
        </reference>
        <reference anchor="HTTP">
          <front>
            <title>HTTP Semantics</title>
            <author fullname="R. Fielding" initials="R." role="editor" surname="Fielding"/>
            <author fullname="M. Nottingham" initials="M." role="editor" surname="Nottingham"/>
            <author fullname="J. Reschke" initials="J." role="editor" surname="Reschke"/>
            <date month="June" year="2022"/>
            <abstract>
              <t>The Hypertext Transfer Protocol (HTTP) is a stateless application-level protocol for distributed, collaborative, hypertext information systems. This document describes the overall architecture of HTTP, establishes common terminology, and defines aspects of the protocol that are shared by all versions. In this definition are core protocol elements, extensibility mechanisms, and the "http" and "https" Uniform Resource Identifier (URI) schemes.</t>
              <t>This document updates RFC 3864 and obsoletes RFCs 2818, 7231, 7232, 7233, 7235, 7538, 7615, 7694, and portions of 7230.</t>
            </abstract>
          </front>
          <seriesInfo name="STD" value="97"/>
          <seriesInfo name="RFC" value="9110"/>
          <seriesInfo name="DOI" value="10.17487/RFC9110"/>
        </reference>
      </references>
    </references>
    <?line 281?>

<section numbered="false" anchor="acknowledgments">
      <name>Acknowledgments</name>
      <t>The author is grateful to the Working Group's chairs, Mark Nottingham
(Cloudflare) and Suresh Krishnan (Cisco), to the editors of the
vocabulary draft, Martin Thomson (Mozilla) and Paul Keller (Open
Future), and to the responsible Area Director, Mike Bishop (Akamai),
whose incredible effort to facilitate discussion and endless patience
(in addition to their own individual contributions) have made the
Working Group's progress possible.</t>
      <t>The author also thanks the following people for their substantial
contributions to shaping the Working Group's discussion and for their
partnership.  Affiliations are listed for identification only, since
IETF participants contribute as individuals.</t>
      <t>Acknowledgement in this section should not be taken to imply
endorsement of, or contribution to, this individual draft; rather, it
recognizes participation in the Working Group, whose work this draft
directly builds on.  Some participants have asked to be excluded from
this list.</t>
      <table>
        <thead>
          <tr>
            <th align="left">Name</th>
            <th align="left">Affiliation</th>
          </tr>
        </thead>
        <tbody>
          <tr>
            <td align="left">Farzaneh Badiei</td>
            <td align="left">Digital Medusa</td>
          </tr>
          <tr>
            <td align="left">Lila Bailey</td>
            <td align="left">Internet Archive</td>
          </tr>
          <tr>
            <td align="left">Kevin Bankston</td>
            <td align="left">Center for Democracy and Technology</td>
          </tr>
          <tr>
            <td align="left">Fabrice Canel</td>
            <td align="left">Microsoft</td>
          </tr>
          <tr>
            <td align="left">Alissa Cooper</td>
            <td align="left">Knight-Georgetown Institute</td>
          </tr>
          <tr>
            <td align="left">Sonia Cooper</td>
            <td align="left">Microsoft</td>
          </tr>
          <tr>
            <td align="left">Glenn Deen</td>
            <td align="left">Comcast NBCUniversal</td>
          </tr>
          <tr>
            <td align="left">Caleb Donaldson</td>
            <td align="left">Google</td>
          </tr>
          <tr>
            <td align="left">Daniel Etcovitch</td>
            <td align="left">Anthropic</td>
          </tr>
          <tr>
            <td align="left">Christopher Flammang</td>
            <td align="left">Elsevier</td>
          </tr>
          <tr>
            <td align="left">Jordan Gimbel</td>
            <td align="left">Microsoft</td>
          </tr>
          <tr>
            <td align="left">Mark Gray</td>
            <td align="left">OpenAI</td>
          </tr>
          <tr>
            <td align="left">Nate Hake</td>
            <td align="left">Travel Lemming</td>
          </tr>
          <tr>
            <td align="left">Gary Illyes</td>
            <td align="left">Google</td>
          </tr>
          <tr>
            <td align="left">Meredith Jacob</td>
            <td align="left">American University Washington College of Law</td>
          </tr>
          <tr>
            <td align="left">Mirja Kühlewind</td>
            <td align="left">Ericsson</td>
          </tr>
          <tr>
            <td align="left">Laurent Le Meur</td>
            <td align="left">EDRLab</td>
          </tr>
          <tr>
            <td align="left">Jo Levy</td>
            <td align="left">Norton Law</td>
          </tr>
          <tr>
            <td align="left">Greg Lindahl</td>
            <td align="left">Common Crawl Foundation</td>
          </tr>
          <tr>
            <td align="left">Krishna Madhavan</td>
            <td align="left">Microsoft</td>
          </tr>
          <tr>
            <td align="left">Sarah McKenna</td>
            <td align="left">Sequentum</td>
          </tr>
          <tr>
            <td align="left">John Mueller</td>
            <td align="left">Google</td>
          </tr>
          <tr>
            <td align="left">Chris Needham</td>
            <td align="left">BBC</td>
          </tr>
          <tr>
            <td align="left">Elaine Newton</td>
            <td align="left">Amazon Web Services</td>
          </tr>
          <tr>
            <td align="left">Victoria Noble</td>
            <td align="left">Electronic Frontier Foundation</td>
          </tr>
          <tr>
            <td align="left">Sebastian Posth</td>
            <td align="left">Liccium</td>
          </tr>
          <tr>
            <td align="left">Brendan Quinn</td>
            <td align="left">IPTC</td>
          </tr>
          <tr>
            <td align="left">Courtney Radsch</td>
            <td align="left">Open Markets Institute</td>
          </tr>
          <tr>
            <td align="left">Felix Reda</td>
            <td align="left">GitHub</td>
          </tr>
          <tr>
            <td align="left">Eric Rescorla</td>
            <td align="left">Independent</td>
          </tr>
          <tr>
            <td align="left">Matt Rogerson</td>
            <td align="left">Financial Times</td>
          </tr>
          <tr>
            <td align="left">Leah Romm</td>
            <td align="left">Cloudflare</td>
          </tr>
          <tr>
            <td align="left">Leonard Rosenthol</td>
            <td align="left">Adobe</td>
          </tr>
          <tr>
            <td align="left">Malikat Rufai</td>
            <td align="left">Meta</td>
          </tr>
          <tr>
            <td align="left">Achim Schlosser</td>
            <td align="left">Bertelsmann</td>
          </tr>
          <tr>
            <td align="left">Rony Shalit</td>
            <td align="left">Bright Data</td>
          </tr>
          <tr>
            <td align="left">Bradley Silver</td>
            <td align="left">Advance</td>
          </tr>
          <tr>
            <td align="left">Erin Simon</td>
            <td align="left">Google</td>
          </tr>
          <tr>
            <td align="left">Esther Tetruashvily</td>
            <td align="left">OpenAI</td>
          </tr>
          <tr>
            <td align="left">Michael Tremante</td>
            <td align="left">Cloudflare</td>
          </tr>
          <tr>
            <td align="left">Thom Vaughan</td>
            <td align="left">Common Crawl Foundation</td>
          </tr>
          <tr>
            <td align="left">James Whymark</td>
            <td align="left">Meta</td>
          </tr>
          <tr>
            <td align="left">Timid Robot Zehta</td>
            <td align="left">Creative Commons</td>
          </tr>
        </tbody>
      </table>
      <t>Many more people shaped the Working Group's discussion than appear
above, and this list has almost certainly left out someone who belongs
on it.  Any such omission is inadvertent, and the author apologizes in
advance.</t>
    </section>
  </back>
  <!-- ##markdown-source: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-->

</rfc>
