Who owns a report, design, software component, or other content that an employee creates with generative AI? What happens when an organization supplies data to train an AI model? And how do employment agreements, intellectual property rights, and data provenance affect the answer?
Arkansas has taken a legislative step toward answering those questions.
Arkansas Act 927 of 2025, originally introduced as HB 1876 and codified at Arkansas Code § 18-4-101, establishes rules for ownership of AI-generated content and trained AI models. The law addresses the person supplying the prompt or training input, employer-directed AI work, contractual transfers, lawful training data, and existing intellectual property rights.
For organizations developing or deploying AI, the practical challenge extends beyond identifying a legal owner. Teams need to know where training data came from, which information entered AI workflows, who directed the work, what contracts govern the resulting assets, and whether the organization can substantiate its position.
This guide explains the Arkansas AI model and generated content ownership law, its limitations, and the steps enterprises can take to strengthen data and AI governance.
Arkansas Act 927 at a Glance
AI Ownership Starts With Inputs, Employment, Contracts, and Existing Rights.
Generated Content
The person providing the input or directive receives the law’s ownership designation, subject to exceptions and IP limits.
Trained Models
The person supplying training data or input may own the resulting trained model if statutory conditions apply.
Employer Direction
Employers receive ownership under the law when qualifying AI work occurs within employment duties and under their direction and control.
IP Boundaries
The law does not grant ownership over content that infringes pre-existing copyright or other intellectual property rights.
The enterprise takeaway: Establish a reliable record of AI inputs, training data provenance, employee direction, and contractual rights before an ownership dispute arises.
What Is the Arkansas AI Model & Generated Content Ownership Law?
Arkansas Act 927 establishes ownership rules for certain activities involving generative artificial intelligence tools. The Arkansas General Assembly enacted the measure in 2025, and the governor approved it on April 21, 2025.
La loi apparaît dans Arkansas Code § 18-4-101, under the state’s property laws.
It addresses three core situations:
- A person generates content: The statute assigns ownership to the person who provides the input or directive, provided the content does not infringe existing copyrights or intellectual property rights.
- A person trains an AI model: The statute assigns ownership of the resulting trained model to the person who provides training data or input, provided that person lawfully acquired the training data and has not transferred ownership through a contract or agreement.
- An employee performs qualifying AI work: The statute assigns ownership to the employer when the employee acts within the scope of employment and under the employer’s direction and control.
The statute also states that it does not grant ownership over content that infringes pre-existing copyrights or other intellectual property rights.
Distinction importante : Act 927 addresses ownership under Arkansas law. It does not eliminate federal copyright requirements, override third-party intellectual property rights, or establish that every AI-generated output qualifies for copyright protection.
Organizations should review the official text of Act 927 and seek legal advice for specific ownership questions.
When Did Arkansas Act 927 Become Law?
The Arkansas legislature approved Act 927 on April 21, 2025. The measure originated as HB 1876 during the 2025 regular session.
By October 2026, the law appears in the Arkansas Code as § 18-4-101.
For an authoritative legislative history, see the Arkansas General Assembly’s HB 1876 record.
Organizations assessing a particular transaction or dispute should confirm the applicable effective date and any relevant contractual or choice-of-law provisions with counsel.
Who Owns AI-Generated Content Under Arkansas Act 927?
Under § 18-4-101(a)(1), the person who supplies the input or directive to a generative AI tool owns the resulting content under the statute, provided the content does not infringe existing copyrights or other intellectual property rights.
That rule creates a starting point, not a universal answer for every AI workflow.
Consider a consultant who uses a generative AI application to draft a presentation. The consultant provides the instructions, but a client agreement may also address ownership, assignment, or licensing of deliverables. If the output reproduces protected third-party material, the Arkansas statute does not erase those rights.
Organizations should therefore examine the full context:
- Who supplied the input or directive?
- Was the person acting as an employee, contractor, or independent user?
- Did an employer direct and control the activity?
- What rights do applicable contracts and platform terms establish?
- Does the output contain or infringe protected third-party material?
What businesses should do: Record the responsible user, business purpose, relevant instructions, approved AI tool, output location, and governing agreements for important AI-generated assets.
Who Owns an AI Model Trained With Enterprise Data?
Arkansas § 18-4-101(a)(2) addresses the person who provides data or input to train a generative AI model.
The statute assigns ownership of the resulting trained model to that person when two conditions hold:
- The person lawfully acquired the training data.
- The person has not transferred ownership rights through a contract or agreement.
This provision creates important questions for organizations that fine-tune models, develop proprietary AI systems, or work with external model developers.
For example, a business might provide customer records, internal documentation, and proprietary research to support a model training project. Before claiming ownership of the resulting model, it needs to examine whether it lawfully acquired and may use that information for the intended training purpose, which party supplied the relevant inputs, and what the vendor agreement says.
Supplying data does not automatically resolve every model ownership dispute. The statute’s conditions, contract terms, other applicable laws, and the facts of the development process all matter.
When Does an Employer Own AI-Generated Content or Model Work?
Arkansas § 18-4-101(b) addresses employees who use generative AI tools as part of their work.
The law assigns the resulting model training data and generated content to the employer when the employee performs the activity:
- Within the scope of employment; and
- Under the direction and control of the employer.
Both statutory conditions matter.
An employee using an approved AI tool to prepare a company report under a manager’s direction presents a different fact pattern from an employee independently using a personal AI account for an unrelated project.
The statute does not provide a complete framework for every contractor, agency, joint-development, or third-party vendor relationship. Those arrangements may depend heavily on contract terms and other applicable law.
The Ownership Decision
Four Questions That Shape the Ownership Analysis
01 / INPUT
Who Supplied It?
Identify the person supplying the content directive or model training input.
02 / EMPLOYMENT
Who Directed It?
Examine employment scope, direction, and control.
03 / CONTRACTS
What Rights Changed?
Review assignments, licenses, vendor terms, and development agreements.
04 / IP RIGHTS
Was the Data Lawful?
Review training data provenance and potential third-party rights.
La principale distinction : An ownership designation under state law does not automatically establish federal copyright eligibility or eliminate competing intellectual property claims.
Does Arkansas Act 927 Give Copyright Protection to AI-Generated Content?
Pas automatiquement.
Arkansas Act 927 addresses ownership of AI-generated content under state law. Federal copyright law asks a separate question: Does the work contain enough human authorship to qualify for copyright protection?
In its January 2025 report on AI and copyrightability, the U.S. Copyright Office explained that copyright may protect human-authored expressive elements in AI-assisted works, but not purely AI-generated material lacking sufficient human authorship.
The Copyright Office also concluded that providing prompts alone does not necessarily establish the human authorship required for copyright protection.
This distinction creates two different questions for enterprises:
- Possession: Which person or organization has a claim to the asset under applicable state law, contracts, and other legal rules?
- Copyrightability: Does the work contain qualifying human expression that federal copyright law protects?
Organizations should avoid treating those questions as interchangeable.
A company may have contractual or state-law rights relating to an AI-generated asset without necessarily holding a federal copyright in every part of that asset.
What Does the Arkansas AI Ownership Law Mean for Enterprise AI Governance?
Act 927 focuses on ownership rather than establishing a broad AI security or privacy compliance program. It does not prescribe a general AI risk assessment process, mandate a particular data discovery technology, or create a universal AI inventory requirement.
However, its ownership conditions make several governance practices operationally valuable.
1. Know Which Data Enters AI Training and Generation Workflows
Ownership questions become difficult when an organization cannot identify the source of its training data or the information employees supplied to AI tools.
Data discovery and classification help teams identify sensitive, personal, confidential, and proprietary information across supported systems.
What to look for: Visibility into relevant data repositories, business-sensitive categories, ownership context, and approved AI data uses.
2. Document Training Data Provenance and Usage Rights
The statute conditions trained-model ownership on lawful acquisition of training data and the absence of a contractual transfer of ownership.
Organizations should maintain evidence about dataset origins, licenses, permissions, restrictions, and the parties involved in model development.
What to look for: Records connecting training datasets to their sources, applicable agreements, and approval decisions.
3. Clarify Employee and Contractor Responsibilities
The employer provision depends on employment scope, direction, and control.
Organizations should establish clear policies for employee AI use and review contracts governing external developers, agencies, and consultants.
What to look for: Documented roles, authorized tools, approved workflows, relevant agreements, and accountability for resulting assets.
4. Assess Sensitive Data and Intellectual Property Exposure
Training data may contain personal information, confidential business records, trade secrets, licensed materials, or copyrighted works.
The ability to identify these categories can help legal, security, and AI teams investigate whether the organization has appropriate rights to use the data.
What to look for: Accurate classification, data ownership context, access governance, and review processes for sensitive AI inputs.
5. Connect AI Assets to Responsible Owners
Organizations benefit from knowing which teams operate AI systems, which datasets support them, and who can approve changes.
These records help teams coordinate legal reviews, security assessments, and governance decisions.
What to look for: An inventory of relevant AI systems and agents, accountable owners, data relationships, and approval workflows.
Five Enterprise Scenarios That Illustrate the Law
The following scenarios illustrate questions organizations may encounter. They do not predict how a court would resolve a specific dispute.
| Scénario | Ownership Question | Evidence to Review |
|---|---|---|
| Employee-generated report | Did the employee act within job duties under employer direction and control? | Work assignment, AI policy, tool records, employment terms |
| Fine-tuned enterprise model | Who supplied training inputs, and did the organization lawfully acquire the data? | Dataset provenance, licenses, development contracts |
| Agency-created marketing assets | What rights did the agency and client establish? | Statement of work, assignment clauses, platform terms |
| AI-generated software component | Who directed creation, and does the output raise third-party IP concerns? | Development records, code review, licenses, employment terms |
| External model development | Which party supplied data, controlled training, and received rights under the agreement? | Vendor contract, data rights, model deliverables, training records |
How BigID Supports Data and AI Governance Related to Act 927
Arkansas Act 927 does not require organizations to purchase a particular technology. But enterprises assessing AI ownership need dependable information about their data, AI systems, and access relationships.
BigID helps organizations connect data discovery, classification, security, privacy, and AI governance across supported environments.
That foundation can support the factual investigation and governance processes surrounding AI ownership questions.
Connecter les points entre les données et l'IA
Better AI Ownership Decisions Start With Better Data Context.
BIGID DATA INTELLIGENCE
Discover + Classify + Connect Context
Sensitive Data | Data Owners | Access | AI Systems | Governance
↓
Identifier les données sensibles
Discover and classify information relevant to AI use and training decisions.
Examine Access
Understand who or what can reach sensitive enterprise information.
Govern AI Data Risk
Connect supported AI systems and agents with relevant data context.
Support Governance
Help teams coordinate data security, privacy, and policy decisions.
Scope note: BigID can support relevant data and AI governance workflows. It does not independently determine legal ownership, establish copyrightability, or replace contract review and legal counsel.
Discover and Classify Data Relevant to AI Development
BigID's découverte et classification des données capabilities help organizations identify sensitive and business-critical information across supported structured and unstructured sources.
Valeur commerciale : Give legal, security, and AI teams more reliable context when assessing what information an organization may use in AI development.
Understand Sensitive Access and Data Exposure
BigID's Gouvernance de l'accès aux données et DSPM capabilities help organizations examine sensitive data exposure and permissions across supported systems.
Valeur commerciale : Identify unnecessary access to confidential information that could enter AI workflows without appropriate review.
Connect Data Intelligence With AI Governance
BigID's capacités de gouvernance de l'IA help organizations examine relevant AI assets, data relationships, and risk across supported environments.
Valeur commerciale : Support more informed decisions about AI use, data exposure, ownership responsibilities, and governance processes.
Coordinate Privacy and Lifecycle Decisions
BigID connects data intelligence with privacy and compliance workflows that can support data rights, retention, and disposition decisions.
Valeur commerciale : Help teams address privacy and lifecycle obligations alongside AI development and security requirements.
Arkansas Act 927: A Practical Enterprise Readiness Checklist
The statute does not prescribe the following checklist. These steps represent governance practices organizations can consider when assessing AI ownership and related data risks.
Enterprise Readiness
Seven Actions to Strengthen AI Ownership Governance
- Inventory important AI use cases. Identify models, tools, business owners, and high-value generated assets.
- Identify training data sources. Record relevant datasets, origin, sensitivity, and intended use.
- Review data rights. Assess acquisition, licensing, consent, contractual restrictions, and third-party IP issues with counsel.
- Clarify employee AI policies. Document approved tools, employment responsibilities, direction, and control.
- Review vendor and contractor agreements. Examine model rights, output ownership, assignments, licenses, and permitted data use.
- Protect sensitive AI inputs. Apply appropriate access governance, data security, and privacy controls.
- Maintain decision records. Preserve evidence that supports ownership reviews, approvals, and corrective action.
What Should Legal, Security, and AI Leaders Ask Next?
Organizations should pressure-test their governance processes with concrete questions:
- Can we identify the person or team that supplied inputs for important AI-generated assets?
- Can we document the source and lawful acquisition of data used in model training?
- Do our employment policies address AI work within assigned duties and under company direction?
- Do our vendor and contractor agreements clearly address AI outputs and model ownership?
- Can we identify sensitive or proprietary information that employees may expose to AI tools?
- Do we distinguish state-law ownership from federal copyrightability?
- Can we produce reliable evidence when a legal, security, or governance team needs to investigate an AI asset?
These questions matter because organizations cannot resolve ownership disputes through policy statements alone. They need reliable facts about how an asset came into existence and which rights govern it.
The Bigger Lesson: AI Ownership Depends on Data Accountability
Arkansas Act 927 gives organizations a clearer state-law starting point for certain AI ownership questions. It connects ownership to the person supplying a directive or training input, while accounting for employer-directed work, contractual transfers, lawful training data, and existing intellectual property rights.
But the statute does not eliminate the operational work behind those questions.
Organizations still need to understand which data they use, how AI systems access that data, who directs development, which contracts apply, and whether the resulting content raises third-party rights concerns.
BigID helps enterprises establish the data intelligence needed to support those decisions across security, privacy, and AI governance.
The result is a more informed approach to AI development: one that connects innovation with accountability for the information behind it.
Arkansas AI Ownership Law: Frequently Asked Questions
What is Arkansas Act 927?
Arkansas Act 927 of 2025, codified at Arkansas Code § 18-4-101, establishes ownership rules for content generated through generative AI tools and models trained with supplied data or input. It also addresses qualifying employer-directed work and existing intellectual property rights.
Is Arkansas HB 1876 the same as Act 927?
Yes. HB 1876 became Act 927 during Arkansas’s 2025 legislative session. The resulting law appears at Arkansas Code § 18-4-101.
Who owns AI-generated content under Arkansas law?
Under § 18-4-101(a)(1), the person supplying the input or directive owns the resulting content under the statute, provided the content does not infringe existing copyrights or other intellectual property rights. The employer provision and other applicable legal rules may affect the analysis.
Who owns an AI model trained with supplied data?
Under § 18-4-101(a)(2), the person supplying data or input for model training owns the resulting trained model if the person lawfully acquired the training data and has not transferred ownership through a contract or agreement, subject to the statute’s employer provision.
Does an employer own AI content created by an employee?
The statute assigns ownership to the employer when an employee performs qualifying AI work within the scope of employment and under the employer’s direction and control.
Does Arkansas Act 927 override copyright law?
No. The statute does not grant ownership over content that infringes pre-existing copyright or other intellectual property rights. Federal copyrightability remains a separate legal question that generally depends on sufficient human authorship.
Does a prompt automatically make AI-generated content copyrightable?
No. The U.S. Copyright Office has explained that prompts alone do not necessarily establish the human authorship required for federal copyright protection. Copyrightability depends on the human creative contribution to the work.
Does Act 927 require companies to maintain an AI inventory?
No. The statute establishes ownership rules rather than a general AI inventory requirement. However, maintaining records of AI systems, training inputs, responsible personnel, and relevant contracts can support ownership reviews.
Does the law apply to AI training data?
Yes, the statute addresses ownership of a resulting trained model when a person supplies data or input for model training. It conditions that ownership on lawful acquisition of the training data and the absence of a contractual transfer of ownership rights.
How can BigID help organizations respond to AI ownership questions?
BigID helps organizations discover and classify sensitive data, examine access and exposure, and connect data intelligence with security, privacy, and AI governance. These capabilities can support factual investigations and governance decisions, but BigID does not independently determine legal ownership or copyrightability.
Connecter les points entre les données et l'IA
Know Your Data. Strengthen Your AI Governance.
Discover sensitive information, examine data access, and connect security, privacy, and AI governance with BigID.
Sources et lectures complémentaires
- Arkansas General Assembly: Act 927 of 2025, Official Text
- Arkansas General Assembly: HB 1876 Legislative History
- Arkansas Code § 18-4-101
- U.S. Copyright Office: Copyright and Artificial Intelligence
- U.S. Copyright Office: AI Copyrightability Report, January 2025
- BigID: Data Discovery and Classification
- BigID : Gouvernance de l’accès aux données
- BigID : Gestion de la posture de sécurité des données
- BigID : Gouvernance de l’IA
Editorial note: This article reflects publicly available legal and regulatory information as of October 21, 2026. It provides general information, not legal advice. Arkansas Act 927 addresses state-law ownership questions but does not resolve every issue involving federal copyright, contracts, third-party intellectual property, or AI training data rights. Organizations should consult qualified counsel about specific circumstances.
Auteur
