Namespace

A licence a machine cannot read is a licence only a lawyer can enforce.

Every term in the four instruments has a machine-readable name here, so a scanner, an SBOM tool or an agent can tell what a work permits without a human reading sixteen sections. Publishing it changes nothing legally. The vocabulary is non-normative, existing Creative Commons parsers do not understand OpenRoots on their own, and the legalcode always governs.

Terms: 24
Classes: 4
Mapped to CC: 5
Conditional: 4

Serializations

Fetch it in whichever form your tooling reads.

Turtle

/ns.ttl

text/turtle

JSON-LD

/ns.jsonld

application/ld+json

RDF/XML

/ns.rdf

application/rdf+xml

Namespace IRI

https://openroots.org/ns#

The addition

One obligation type nobody else had a name for.

A Creative Commons Requirement is unconditional. Attribution is owed by everyone, always. An OpenRoots obligation can sit dormant and wake up when a specific thing happens, such as crossing the revenue threshold or starting to train a model. The condition that wakes it is part of the term itself rather than a footnote to it, which is what lets a machine evaluate it.

or:RoyaltyPaymentsection 5.1

Pay the stated royalty on attributable revenue, subject to the annual cap.

Activates when

The Licensee is a Canopy Licensee making Production Use.

or:CanopyReportsection 5.2

Submit the quarterly revenue and remittance statement required by the instrument.

Activates when

The Licensee is a Canopy Licensee.

or:ComputeLicencesection 6.2

Obtain a separate Compute licence and pay the metric-based fee.

Activates when

The Licensee makes AI Training Use.

or:TrainingOutputProvenancesection 6.3

Provide the required notice when an output reproduces an Extracted Fragment above the Exhibit B threshold.

Activates when

A training output contains a qualifying Extracted Fragment.

Permission

What the instrument allows you to do without asking.

Term	Section	Meaning	CC equivalent
or:Use	2.1	Use the Work for permitted purposes, subject to the instrument's conditions.	none
or:Reproduction	2.1	Make copies of the Work in a form covered by the artifact-specific grant.	cc:Reproduction
or:Distribution	2.1	Distribute the Work, subject to Sections 4, 6, 8, and 10.	cc:Distribution
or:Modification	2.1	Modify the Work and exercise the artifact-specific derivative rights in Section 2.1.	cc:DerivativeWorks
or:ProductionUse	2.1	Make Production Use under the Root or Canopy grant; Section 5 may attach for a Canopy Licensee.	none
or:InternalProduction	4.2	Operate the Work internally at production scope under the Section 4.2 carve-out.	none
or:LargerProduct	4.2	Incorporate the Work into a larger product with substantial independent functionality, as defined by the applicable Section 4.2 carve-out.	none
or:ProfessionalServices	4.2	Provide qualifying services or customer-controlled deployments under the applicable Section 4.2 carve-out.	none
or:MarketplaceListing	4.2	List qualifying works on a multi-provider platform under the applicable Section 4.2 carve-out.	none

Prohibition

What the instrument does not allow at any tier.

Term	Section	Meaning	CC equivalent
or:CompetingCommercialization	4.1	Do not commercialize the Work or a Derivative Work as a Competing Offering without a separate written agreement.	none
or:WorkAsProduct	4.3	Root status does not permit commercializing the Work itself through resale, repackaging, porting, hosting, or redistribution.	none
or:UnlicensedAITraining	6.1	Do not make AI Training Use without a separate Compute licence.	none
or:FalseFallbackClaim	7.3	Do not claim that time, a review date, silence, or non-enforcement creates rights under another licence.	none
or:PatentAggression	9.2	A patent claim alleging that the Work infringes terminates the patent licence as Section 9.2 provides.	none
or:MisleadingTrademarkUse	11.1	Do not imply endorsement or apply an OpenRoots instrument name to altered terms.	none

Requirement

What you owe from the moment you use the artifact.

Term	Section	Meaning	CC equivalent
or:NoticePreservation	8.1	Retain the copyright notice and the licence copy or reference required by the artifact-specific Section 8.1.	cc:Notice
or:LicensorAttribution	8.1	Identify the original Licensor as required by the artifact-specific Section 8.1.	cc:Attribution
or:VersionDisclosure	8.1	Identify the applicable OpenRoots instrument version on covered copies or distributions.	none
or:ContributionIntegrity	8.2	Identify original, licensed, or AI-assisted contributions and reports as the artifact-specific Section 8 requires.	none
or:ChangeDisclosure	8.2	Provide the artifact-specific modification, provenance, or compatibility disclosure required by Section 8.	none

AI preferences

What goes in the standard header, and what stays ours.

The active IETF draft defines train-ai and search. It records a preference and creates no rights. OpenRoots emits only those two registered categories in Content-Usage, then publishes its own finer legal concepts separately, so a private extension is never dressed up as an IETF standard.

Category	Source	Definition
train-ai	IETF draft-07	Using an asset to modify learned parameters of an AI model used to generate synthetic content in one or more modalities.
search	IETF draft-07	Using assets in a search application that directs users back to the location the asset came from. Output carries a reference to that location and may quote only excerpts drawn directly from the asset.

What a licensor publishes

Content-Usage: train-ai=y, search=y
OpenRoots-Terms: instrument=ORD-2.2; credit=required; provenance=required; threshold=20000000; rate=0.005; cap=250000; disclosure=model-card; ai-training=separate-compute-licence; competing-offering=separate-written-agreement; competing-offering-scope=all-licensees; fallback=none; conversion-months=none
Link: <https://openroots.org/licenses/ord/2.2/>; rel="license"

OpenRoots concept	IETF mapping	Section	Legal effect
or:train-genai Training or fine-tuning a model whose principal purpose is generating new content in one or more media.	train-ai	1.10	OpenRoots classifies this as AI Training Use, so Section 6 requires a separate Compute licence.
or:train-distill Using the outputs, logits, labels, or other behavior of one model as training input or supervision for another model.	train-ai	1.10	Distillation is named expressly in the AI Training Use definition and triggers Section 6.
or:rag-index Building or using a retrieval index over an artifact so selected material is supplied to a model at inference or evaluation time.	search	1.10	Ordinary search remains permitted by Section 6.4, but retrieval augmentation is named as AI Training Use and requires the Section 6 Compute licence. The IETF preference does not grant or override legal rights.

OpenRoots concept

IETF mapping

Section

Legal effect

or:train-genai

Training or fine-tuning a model whose principal purpose is generating new content in one or more media.

train-ai

1.10

OpenRoots classifies this as AI Training Use, so Section 6 requires a separate Compute licence.

or:train-distill

Using the outputs, logits, labels, or other behavior of one model as training input or supervision for another model.

train-ai

1.10

Distillation is named expressly in the AI Training Use definition and triggers Section 6.

or:rag-index

Building or using a retrieval index over an artifact so selected material is supplied to a model at inference or evaluation time.

1.10

Ordinary search remains permitted by Section 6.4, but retrieval augmentation is named as AI Training Use and requires the Section 6 Compute licence. The IETF preference does not grant or override legal rights.

These three identifiers are OpenRoots vocabulary. Unknown Content-Usage labels may be ignored by conforming implementations, so none of them appears in that header.

CC signals

Where OpenRoots goes further than the CC signals, and where it declines.

The four Creative Commons signal elements are a comparison framework rather than incorporated terms. OpenRoots matches one duty, quantifies a second on a different trigger, and declines two of them on purpose.

Credit

same

You must give appropriate credit based on the method, means, and context of your use.

Comparable, but instrument-specific. Section 8.1 requires the copyright notice, licence reference, original Licensor, and applicable version on covered copies or distributions.

Direct Contribution

stronger

You must provide monetary or in-kind support to the Declaring Party, based on a good faith valuation taking into account your use and your financial means.

Quantified rather than good faith. Section 5.1 sets the Canopy royalty and cap; Section 6.2 separately requires a metric-based Compute Fee for AI Training Use. They are different triggers, not one general contribution duty.

Ecosystem Contribution

declined

You must provide monetary or in-kind support back to the ecosystem from which you are benefiting.

Not adopted. Payment runs to the named licensor, never to an undefined ecosystem, because an obligation owed to nobody in particular cannot be enforced or discharged.

Open

declined

The AI system used must be open, satisfying MOF Class II, MOF Class I, or the OSAID.

Not adopted. OpenRoots does not require model openness. Artifact-specific Section 8 disclosures and Section 6.3 output provenance are narrower obligations and should not be represented as an openness requirement.

The full reasoning behind each of these is in the decision log on stewardship.