AI Governance Vocabulary Development Develop and promote an interoperable ODRL profile for expressing machine-readable policies governing the use of assets in artificial intelligence workflows. Submitter's Note: This plan has been inferred by ChatGPT from the W3C ODRL Community Group's ODRL Profile: AI Vocabulary early draft, the purposes of the ODRL Community Group ^^ It was prompted by a LinkedIn posting by Renato Iannella and comment by Leonard Rosenthol: https://www.linkedin.com/posts/riannella_the-w3c-odrl-community-group-has-released-activity-7480448082665500672-s_j-?utm_source=share&utm_medium=member_desktop&rcm=ACoAAACRo-wBsB6AJXaqtz06r_wMwIYVUJtr0PM ^^ According to ChatGPT: The profile defines AI-specific actions through which policy authors can express acceptable and prohibited uses of assets in AI workflows. It is intended to support governance, risk management, and compliance by helping AI systems operate within agreed legal, ethical, organizational, and contractual boundaries. ^^ The ODRL AI Vocabulary and the IETF AI Preferences effort appear to be complementary. The IETF effort can standardize mechanisms and vocabulary through which creators, publishers, and other rights holders communicate preferences concerning AI uses of content. The ODRL profile can express more detailed permissions, prohibitions, duties, parties, assets, actions, and constraints governing such uses. Coordination could reduce semantic conflicts, duplication, and incompatible implementations. ^^ StratML complements both efforts by documenting the intentions, stakeholders, objectives, and prospective results of the organizations developing and implementing the standards. ODRL Community Group ODRLCG e505df1a-6269-4e7b-b18d-a0266de4b847 Supports the adoption and promotion of the W3C ODRL Recommendations and their continual evolution and future development. World Wide Web Consortium Hosts the ODRL Community Group and provides the standards-development environment in which ODRL recommendations, profiles, reports, and related activities are developed and promoted. ODRL Community Group Participants Contribute use cases, requirements, vocabulary terms, technical expertise, reviews, implementations, and consensus to the development and evolution of ODRL. Renato Iannella Serves as a leader of the ODRL Community Group and promotes review, adoption, coordination, and continuing development of ODRL specifications and profiles. Michael Steidl Contributes leadership relating to ODRL profiles. Víctor Rodríguez-Doncel Contributes leadership relating to the ODRL developer community. David Filip Contributes leadership relating to industry outreach. W3C Members Support, review, implement, and contribute to Web standards and related community activities. Web Users Depend upon transparent, interoperable, trustworthy, and rights-respecting uses of assets and services on the Web. An environment in which artificial intelligence systems use assets in accordance with transparent, interoperable, machine-readable governance policies. VISION-ODRL-AI To dfevelop and promote an ODRL profile enabling acceptable and prohibited uses of assets in artificial intelligence workflows to be expressed and interpreted consistently. d2046788-0ad1-4cab-bf86-a4800657ff5f Interoperability Enable policies to be exchanged, processed, and interpreted consistently across systems, organizations, sectors, and jurisdictions. Transparency Make applicable permissions, prohibitions, duties, actions, parties, assets, and constraints explicit and discoverable. Accountability Clarify the policies under which parties and AI systems use assets and support evaluation of their conformance. Agency Enable authorized parties to express terms governing AI uses of assets in forms that can be communicated and processed by machines. Responsibility Support AI practices consistent with legal, ethical, organizational, societal, and contractual expectations. Openness Invite public participation, review, feedback, implementation, and refinement of the vocabulary and profile. Extensibility Accommodate evolving AI technologies, use cases, policy requirements, and sector-specific profiles without sacrificing shared semantics. Precision Define AI-related actions and policy concepts clearly enough to minimize ambiguity and inconsistent interpretation. Vocabulary Define a controlled vocabulary of actions governing uses of assets by artificial intelligence systems. dc854001-dfc7-4be3-a85d-9210c852e69a 1 Policy Authors Use vocabulary terms to express permissions, prohibitions, duties, and constraints governing AI uses of assets. Asset Rights Holders Depend upon sufficiently precise terms to express conditions governing the use of assets in AI workflows. AI System Developers Implement the vocabulary in systems that discover, interpret, apply, and communicate AI governance policies. The early draft identifies a need for precise, machine-readable policies addressing AI interactions with digital and physical assets through activities such as training, inference, transformation, generation, and redistribution. The vocabulary should remain aligned with the ODRL Information Model and Vocabulary while adding semantics required for AI governance. Actions Define AI-specific actions representing significant uses of assets in AI workflows. 223fdf1c-40be-4a97-9295-d44fafd2189e 1.1 Vocabulary Editors Draft and maintain definitions, usage notes, semantic relationships, and examples for AI-specific actions. Domain Experts Identify material AI activities and evaluate whether proposed terms accurately represent real-world practices. Relevant activities may include training models, fine-tuning models, performing inference, generating outputs, modifying assets, creating derivatives, extracting information, evaluating systems, and redistributing AI-derived content. Terms should be defined according to demonstrated governance use cases rather than merely cataloging technical operations. Semantics Specify the meanings and relationships of AI governance actions consistently with the ODRL model. fc61b1cb-d918-4408-aec2-f18277d18fe7 1.2 Semantic Web Practitioners Contribute expertise in vocabularies, linked data, ontologies, identifiers, and machine-processable semantics. Policy Processing System Developers Apply the semantics when interpreting and evaluating ODRL policies. Semantic definitions should distinguish actions that may appear similar but have materially different governance implications. Relationships among broader, narrower, equivalent, dependent, and potentially conflicting actions should be explicit where useful for processing and interpretation. Examples Provide representative policy examples demonstrating how the vocabulary can express acceptable and prohibited AI uses. 77d48f32-dbb3-4164-a1d7-e4e163dc3c45 1.3 Implementers Use examples to understand the intended expression and processing of AI governance policies. Policy Practitioners Evaluate whether examples reflect legal, ethical, contractual, and organizational policy requirements. Examples should cover permissions, prohibitions, and duties involving different assets, parties, purposes, technologies, contexts, and constraints. They should illustrate both straightforward and potentially conflicting policy situations. Policies Enable authorized parties to express machine-readable governance policies for AI uses of assets. e05f9ae1-8199-4481-8ccb-22d8de741ce7 2 Content Creators Express conditions governing AI uses of their works and other assets. Publishers Communicate policies governing AI access to and use of published assets. Data Providers Specify permissible and prohibited AI uses of datasets and data services. AI Providers Express and interpret policies applicable to AI models, systems, services, inputs, and outputs. ODRL policies can associate rules with assets and parties and can express permissions, prohibitions, duties, and constraints. The AI profile should make those capabilities practically usable for AI governance scenarios without implying that technical expression alone establishes legal authority or guarantees compliance. Permissions Enable parties to specify AI uses of assets that are allowed under stated conditions. b93ad24d-46f7-4d8a-aa92-0cb43be2d365 2.1 Policy Assigners Declare permissions when they possess or represent authority to authorize specified AI uses. Policy Assignees Interpret permissions and associated conditions before undertaking AI uses of assets. Permissions may be limited by purpose, party, location, time, system, model, security control, attribution requirement, compensation term, or other constraints supported by ODRL or compatible profiles. Prohibitions Enable parties to specify AI uses of assets that are not allowed. 1b6d20d2-9dff-4c97-b365-690b4920c1f9 2.2 Rights Holders Communicate AI uses that they prohibit within the scope of their authority. AI Operators Identify and avoid prohibited uses when applicable policies are authoritative and technically interpretable. Machine-readable prohibitions can improve discoverability and automated handling but do not by themselves resolve questions about legal validity, jurisdiction, exceptions, conflicts, authorization, or enforceability. Duties Enable parties to specify actions that must accompany or follow permitted AI uses. 0b54343e-65e9-4343-a457-b14dff567df3 2.3 Policy Assigners Specify obligations associated with authorized uses of assets. Policy Assignees Determine and perform applicable duties when exercising permissions. Duties may include attribution, notice, reporting, payment, deletion, security controls, provenance preservation, output labeling, audit support, or other actions. Definitions should distinguish duties from constraints and from actions that are themselves the subject of permissions or prohibitions. Constraints Enable policies to limit AI-related permissions, prohibitions, and duties according to relevant circumstances. 3c165788-d430-4108-a475-7934ff63f85d 2.4 Policy Designers Define conditions needed to represent context-sensitive AI governance requirements. Compliance System Developers Implement mechanisms for interpreting and evaluating policy constraints. Constraints may address purposes, parties, asset classes, model types, deployment contexts, jurisdictions, dates, quantities, safeguards, output characteristics, and other factors. The profile should reuse existing ODRL terms and compatible vocabularies where possible rather than creating redundant concepts. Interoperability Promote consistent exchange and interpretation of AI governance policies across technologies, organizations, sectors, and jurisdictions. ee5cd9b4-ade0-4176-b34b-3c6c71ef95c5 3 Standards Organizations Coordinate related specifications, vocabularies, protocols, formats, and implementation guidance. Platform Providers Implement interoperable mechanisms for creating, publishing, discovering, interpreting, and applying policies. Regulators Consider whether machine-readable policies can support regulatory communication, compliance, oversight, and evidence without replacing authoritative legal requirements. Interoperability requires more than shared syntax. Implementations must share sufficiently precise semantics, processing expectations, identifiers, precedence rules, and methods for handling unknown or conflicting terms. Alignment Align the AI vocabulary with the ODRL Information Model, ODRL Vocabulary, profile guidance, and compatible standards. ff217648-bdd3-4b74-97dd-681ab2d0b099 3.1 World Wide Web Consortium Maintains ODRL recommendations and related Web standards with which the profile should remain compatible. ODRL Profile Developers Identify opportunities to reuse terms and maintain compatibility among domain-specific profiles. Alignment should include reuse or mapping of applicable terms from related standards and vocabularies, including those addressing privacy, provenance, rights, licensing, data governance, content authenticity, and artificial intelligence. Coordination Coordinate vocabulary and implementation work with the IETF AI Preferences Working Group and other relevant standards communities. c04d50eb-8ecd-4632-9caf-7eab22994e66 3.2 IETF AI Preferences Working Group Develops interoperable mechanisms and vocabulary for communicating preferences concerning the use of content by artificial intelligence systems. Leonard Rosenthol Highlighted the need for coordination between the ODRL Community Group and the IETF AI Preferences effort. Liaison Participants Communicate requirements, terminology, decisions, dependencies, and potential conflicts among standards-development communities. The IETF AI Preferences effort may provide broadly deployable signals and protocol mechanisms through which content-related AI preferences are communicated. ODRL can provide richer policy expressions involving permissions, prohibitions, duties, parties, assets, and constraints. Coordination should clarify scope, terminology, mappings, precedence, and implementation pathways rather than presuming that either effort can or should replace the other. Mappings Document mappings between AI vocabulary terms and related concepts in other standards and policy languages. 5153bdd8-2e85-4593-ac5a-28285490c9af 3.3 Vocabulary Maintainers Identify semantic correspondences, differences, dependencies, and conflicts among vocabularies. Integration Developers Use mappings to translate or reconcile policy information across systems and formats. Mappings should disclose whether terms are equivalent, broader, narrower, overlapping, merely related, or incompatible. Superficial lexical similarity should not be treated as semantic equivalence. Implementation Support practical creation, publication, discovery, interpretation, evaluation, and application of AI governance policies. a696cad9-5338-4594-bf3c-7540f9f1694c 4 Software Developers Build tools and services supporting ODRL AI policy workflows. AI System Operators Integrate applicable policy discovery, interpretation, decision, documentation, and compliance functions into AI workflows. Auditors Evaluate whether implementations process and apply policies as represented and claimed. End Users Depend upon usable, understandable, and trustworthy tools for expressing and applying AI governance policies. Implementation should preserve the distinction between a policy statement, evidence that a policy applies, authorization by a competent party, interpretation by a processor, and actual conformance by an AI system or operator. Guidance Publish implementation guidance for producing and processing conformant AI governance policies. 3f03750b-ab9b-4275-a89f-22d582296078 4.1 Implementers Apply guidance when creating interoperable policy authoring, publication, discovery, interpretation, and evaluation capabilities. Guidance should address identifiers, profile declarations, serialization, term reuse, validation, unknown terms, policy inheritance, conflict handling, error behavior, security, privacy, provenance, and human-readable presentation. Tools Encourage development of interoperable tools for authoring, validating, publishing, discovering, interpreting, and evaluating policies. 9762b9de-9a85-49e8-ae25-c03a38793d56 4.2 Tool Developers Create software supporting policy authors, publishers, AI providers, operators, auditors, and users. Policy Authors Use tools to create policies accurately without needing to manipulate low-level serializations directly. Tools should make policy meaning visible to users, distinguish machine validation from legal or organizational authority, and avoid implying compliance merely because a document is syntactically valid. Testing Develop test cases and implementation evidence demonstrating consistent processing of the profile. dca8568b-dace-4bbf-8847-ebf8d802cb15 4.3 Test Developers Create conformance cases covering valid, invalid, ambiguous, unsupported, and conflicting policies. Independent Implementers Demonstrate whether separately developed systems interpret and process profile terms consistently. Testing should include positive and negative cases, edge cases, policy conflicts, unsupported terms, multiple profiles, inheritance, constraints, and combinations of permissions, prohibitions, and duties. Interoperability claims should be supported by evidence from independent implementations. Adoption Promote responsible adoption of the profile in relevant AI, data, media, governance, and compliance ecosystems. 43893c56-1af6-42d7-89bc-15b1c5718c40 4.4 Industry Consortia Evaluate and apply the profile in shared AI governance and data-use arrangements. Public Agencies Consider the profile for machine-readable expression of applicable AI usage policies and requirements. Civil Society Organizations Evaluate whether implementations support agency, accountability, transparency, access, fairness, and protection of affected interests. Adoption should be guided by demonstrable use cases and results rather than by publication of the vocabulary alone. Implementers should document what policy signals they recognize, how they interpret them, and what actions their systems take in response. Governance Maintain the profile through open participation, evidence, feedback, and accountable standards-development processes. 6646d6cc-47f4-49dd-98da-4189e9784c2d 5 ODRL Community Group Participants Review proposals, discuss issues, contribute text, evaluate implementations, and seek consensus on profile development. Public Reviewers Identify ambiguities, omissions, conflicts, unintended consequences, and prospective improvements. Affected Communities Provide perspectives on how AI policy expressions may affect rights, opportunities, responsibilities, and risks. The early draft explicitly invites feedback on all aspects of the AI vocabulary. Governance should remain responsive to implementation experience, changes in AI technology, evolving legal and contractual practices, and related standards-development activities. Participation Invite broad and informed participation in reviewing and improving the AI vocabulary and profile. c6aa1e6e-4893-47b6-8878-89456e0a00f0 5.1 AI Governance Practitioners Contribute operational, legal, ethical, contractual, risk-management, and compliance use cases. Researchers Contribute evidence, conceptual analysis, formal methods, evaluations, and emerging use cases. Workshop Participants Help identify future ODRL standardization priorities and requirements. The W3C Workshop on the Future of ODRL, scheduled for July 20–21, 2026, in London and online, provides an opportunity for industry, research, policy, and implementation communities to consider AI governance and other priorities for ODRL's evolution. Review Evaluate proposed terms and specifications through transparent public review and issue resolution. b0778ffe-34a7-4ca8-b6b1-b8a4ff0b2ae4 5.2 Specification Reviewers Assess clarity, completeness, consistency, implementability, interoperability, and potential effects. Editors Document issues, proposals, decisions, revisions, and unresolved questions. Review should distinguish consensus about technical representation from agreement about substantive legal, ethical, contractual, or public-policy rules. The profile should enable policies to be expressed without purporting to determine which policies ought to govern every context. Evolution Revise the vocabulary and profile in response to validated requirements, implementation experience, and changes in related standards. f99906e5-cf5c-4789-8443-f5fb863699d2 5.3 Profile Maintainers Manage additions, corrections, deprecations, versioning, compatibility, and documentation. Implementers Report operational experience and assess the compatibility effects of proposed revisions. Evolution should preserve stable identifiers and semantics where feasible, clearly document changes, and provide migration guidance when revisions affect existing policies or implementations. Accountability Document decisions, responsibilities, implementation claims, and results associated with development and use of the profile. 41841d22-2580-4004-893d-a7662ea29331 5.4 Standards Participants Document contributions, decisions, objections, commitments, and responsibilities through open standards processes. Profile Adopters Disclose supported terms, processing behavior, limitations, and evidence underlying conformance or interoperability claims. Future rendition in StratML Part 2 format could associate quantitative and qualitative performance indicators with objectives such as vocabulary coverage, issue resolution, independent implementations, test-case passage, cross-standard mappings, deployment, policy processing, and documented conformance. 2026-05-27 2026-07-10 https://w3c.github.io/odrl/ai-vocab/ Owen Ambur Owen.Ambur@verizon.net