Machine-Readable Training Preference Signal

| Internet-Draft | Training Preference Signal | October 2026 | | Kelley | Expires 3 May 2027 | [Page] |

This document describes how a service operator binds usage preferences (e.g., "no AI training") to paths by writing ordinary robots.txt groups keyed by a per-purpose policy token in the User-Agent line. It introduces no new directive names and no new grammar: the path language, longest-match precedence, and group structure of robots.txt

are not changed.[¶](#section-abstract-1)

This note is to be removed before publishing as an RFC.[¶](#section-note.1-1)

        Status information for this document may be found at [https://datatracker.ietf.org/doc/draft-kelley-aipref-training-preference-signal/](https://datatracker.ietf.org/doc/draft-kelley-aipref-training-preference-signal/).[¶](#section-note.1-2)

        Discussion of this document takes place on the
        AI Preferences Working Group mailing list ([mailto:ai-control@ietf.org](mailto:ai-control@ietf.org)),
        which is archived at [https://mailarchive.ietf.org/arch/browse/ai-control/](https://mailarchive.ietf.org/arch/browse/ai-control/).
        Subscribe at [https://www.ietf.org/mailman/listinfo/ai-control/](https://www.ietf.org/mailman/listinfo/ai-control/).[¶](#section-note.1-3)

        This Internet-Draft is submitted in full conformance with the
        provisions of BCP 78 and BCP 79.[¶](#section-boilerplate.1-1)

        Internet-Drafts are working documents of the Internet Engineering Task
        Force (IETF). Note that other groups may also distribute working
        documents as Internet-Drafts. The list of current Internet-Drafts is
        at [https://datatracker.ietf.org/drafts/current/](https://datatracker.ietf.org/drafts/current/).[¶](#section-boilerplate.1-2)

        Internet-Drafts are draft documents valid for a maximum of six months
        and may be updated, replaced, or obsoleted by other documents at any
        time. It is inappropriate to use Internet-Drafts as reference
        material or to cite them other than as "work in progress."[¶](#section-boilerplate.1-3)

        This Internet-Draft will expire on 3 May 2027.[¶](#section-boilerplate.1-4)

            Copyright (c) 2026 IETF Trust and the persons identified as the
            document authors. All rights reserved.[¶](#section-boilerplate.2-1)

            This document is subject to BCP 78 and the IETF Trust's Legal
            Provisions Relating to IETF Documents
            ([https://trustee.ietf.org/license-info](https://trustee.ietf.org/license-info)) in effect on the date of
            publication of this document. Please review these documents
            carefully, as they describe your rights and restrictions with
            respect to this document. Code Components extracted from this
            document must include Revised BSD License text as described in
            Section 4.e of the Trust Legal Provisions and are provided without
            warranty as described in the Revised BSD License.[¶](#section-boilerplate.2-2)

A service operator binds usage preferences (e.g., "no AI training") to paths by writing ordinary robots.txt groups keyed by a

per-purpose policy token in the User-Agent line.¶ This expresses a preference against AI training for all paths except those under /press-kit/, for which it expresses a preference

in favor (Section 4.2).¶ This document introduces no new directive names and no new grammar; the path language, longest-match precedence, and group structure defined by [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.¶ Path-specific bindings. Preferences can be expressed per path,

and not only site-wide (Section 4.2).¶ Minimal distance from RFC 9309. Semantics, and ideally

syntax, stay as close to [[ROBOTS](#ROBOTS)] as possible, so that authors and
implementers can use the mental model they already have.[¶](#section-1.1-1.2.1)

No large-scale duplication. robots.txt files are size-constrained (Section 2.5 of [ROBOTS]), and some deployed files already contain heavily duplicated groups. Expressing preferences does not require restating existing rules

(Section 4.1).¶ Unchanged legacy behavior. 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 ([Section 4.2](#checking) and
[Section 4.3](#independence)).[¶](#section-1.1-1.4.1)

Consistent interpretation across consumers. This mechanism does not depend on the divergent parts of deployed parser behavior

(Section 4.1).¶ The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as

described in BCP 14 [[RFC2119](#RFC2119)] [[RFC8174](#RFC8174)] when, and only when, they
appear in all capitals, as shown here.[¶](#section-2-1)

An entity that operates an HTTP service, such as a website, and publishes a robots.txt file for it. For preferences expressed through this mechanism, the service operator is the "declaring

party" ([Section 2](https://datatracker.ietf.org/doc/html/draft-ietf-aipref-vocab-08#section-2) of [[VOCAB](#VOCAB)]).  [[ROBOTS](#ROBOTS)] calls this party a
"service owner".[¶](#section-2-2.2.1)

An entity that interprets preference signals that a service operator

expresses through the mechanism defined in this document.[¶](#section-2-2.4.1)

In the sense of [[ROBOTS](#ROBOTS)]; software that parses `robots.txt` files
and matches product tokens and paths against their rules.[¶](#section-2-2.6.1)

A product token ([Section 2.2.1](https://rfc-editor.org/rfc/rfc9309#section-2.2.1) of [[ROBOTS](#ROBOTS)]) whose RFC 9309 signals

a consumer uses to determine how content will be used. A policy token need not appear in the User-Agent header field

([Section 10.1.5](https://rfc-editor.org/rfc/rfc9110#section-10.1.5) of [[HTTP](#HTTP)]) of any outbound request.  This document
introduces a new, cross-vendor policy token ([Section 4.1](#policy-token)).[¶](#section-2-2.8.1)

A policy token documented by a single consumer. Some service

operators already include these in their robots.txt files.¶ The __usage_ai-training policy token corresponds to the definition that the IETF aipref Working Group has arrived at

([Section 4.1](https://datatracker.ietf.org/doc/html/draft-ietf-aipref-vocab-08#section-4.1) of [[VOCAB](#VOCAB)]).  The following definitions are quoted from
that document:[¶](#section-3-1)

Using an asset to modify the learned parameters of a generative AI
model.[¶](#section-3-2.2.1)

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.¶ 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.[¶](#section-3-2.4.1)

This document applies those definitions as follows:[¶](#section-3-3)

The content retrieved from a path. The preferences that apply to an asset are those expressed in the robots.txt file in effect when the

asset was retrieved (Section 2.4 of [ROBOTS]; see also [ATTACH]).¶ This document does not define a parallel vocabulary; it tracks the work of the aipref Working Group. Implementers SHOULD consider updating their definitions, and adding or removing vocabulary terms,

as the aipref Working Group reaches consensus in additional areas.[¶](#section-3-5)

This document defines the policy token `__usage_ai-training`.[¶](#section-4.1-1)

The policy token conforms to the product-token syntax in

Section 2.2.1 of [ROBOTS] (letters, _, -).¶ Consumers MUST match policy tokens exactly, apart from case, as for any product token. (Note that some deployed parsers match more

flexibly; the __ prefix makes accidental matches unlikely.)¶ A single group MAY name several policy tokens (consecutive User-Agent lines; Section 2.2 of [ROBOTS]) to bind its rules to several purposes. Policy tokens and other product tokens MAY be

mixed in a group.¶ 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 [2.2.1](https://rfc-editor.org/rfc/rfc9309#section-2.2.1) and [2.2.2](https://rfc-editor.org/rfc/rfc9309#section-2.2.2) of [[ROBOTS](#ROBOTS)].[¶](#section-4.2-1)

However, wildcard groups (i.e., `User-Agent: *`) MUST NOT be interpreted
as matching the policy token.[¶](#section-4.2-2)

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.[¶](#section-4.2-3)

If no group contains the `__usage_ai-training` policy token (i.e.,
there is no `User-Agent: __usage_ai-training` line in the file), then

no preference has been expressed. If such a file contains a proprietary policy token, consumers SHOULD interpret that token as they would in the absence of this document, but MUST NOT treat it as expressing a preference in the context of this document (see

Section 4.5).¶ Consumers MUST 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, ExampleBot still checks whether there are rules for ExampleBot; if there are, they still express the service operator's preference about crawling. Nothing in this document

changes that signaling.¶ Content-usage preferences and traffic management signals can both be

expressed in `robots.txt`, but they are independent and orthogonal.[¶](#section-4.3-2)

This policy token is not a directive for crawling behavior.[¶](#section-4.4-1)

A consumer MAY use preferences to inform crawling. For example, if it crawls only for purposes covered by the policy token, it MAY refrain

from retrieving resources for which a preference against those
purposes has been expressed.[¶](#section-4.4-2)

If a service operator wishes to limit crawling, it MAY use groups for
crawlers' product tokens in `robots.txt` to do so.[¶](#section-4.4-3)

Specific expressions SHOULD take precedence over general ones.[¶](#section-4.5-1)

Consumers MAY document that they interpret rules that match proprietary policy tokens as being more specific than the

`__usage_ai-training` policy token.[¶](#section-4.5-2)

Private agreements MAY override preferences expressed via this
mechanism.[¶](#section-4.5-3)

The security considerations of [Section 3](https://rfc-editor.org/rfc/rfc9309#section-3) of [[ROBOTS](#ROBOTS)] apply to this
mechanism.[¶](#section-5-1)

This document does not introduce any new privacy considerations.[¶](#section-6-1)

This document has no IANA actions.[¶](#section-7-1)

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.¶ 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.¶

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.¶