Generative Artificial Intelligence and copyright - legal illustration

⚖️ Intellectual Property & AI

Generative Artificial Intelligence and Intellectual Property: What You Create with AI May Not Be Yours. Three Legal Risks for Businesses and Creators in the Dominican Republic

By Carlos Romero Polanco 05/06/2026 Reading time: 7 min

When a company, independent professional, designer, developer, or consultant deploys AI-generated results, they face three legal issues: legal protection, ownership, and confidentiality. In the Dominican Republic, the lack of specific regulation makes these risks even more critical.

📌 Executive Summary

Generative AI raises three legal questions for businesses and independent creators: first, whether the result has sufficient human intervention to be protected as a work; second, whether the user has a valid title to exploit it; and third, whether the generation process compromised confidential information, trade secrets, personal data, or third-party rights. Law 65-00 reserves authorship to the natural person, but does not prevent a legal entity from exercising economic rights as a derivative owner when a protectable work exists and there is a valid mechanism for attribution or assignment. Laws 20-00 and 172-13 complete the analysis regarding industrial property and data protection. A company does not need to be the author to exercise economic rights; it needs a protectable work and a valid legal title. For the independent creator, the key is different: they must be able to demonstrate sufficient creative human input over the AI-generated result.

           Critical fact: The prompt alone is usually not sufficient creative input. If the system determines the essential expressive elements without sufficient human control, the result may fall outside copyright protection.
        

I. Authorship and Possibility of Legal Protection for AI-Generated Results

Article 5 of Law 65-00 reserves authorship to the natural person. This rule does not prevent a legal entity from exercising economic rights, but that presupposes the existence of a protectable work. Article 7 limits protection to the form of expression and excludes ideas, procedures, and methods. The technical consequence is direct: when a company produces results with AI and the system determines the essential expressive elements without sufficient creative human input, the result may not constitute a protectable work under Law 65-00. In that case, there are no economic rights that can be assigned or acquired.

The U.S. Copyright Office clarified in its January 2025 report: the prompt alone is usually not sufficient creative input; the result without sufficient human control is not protected. However, the selection, coordination, arrangement, or creative modification that a natural person makes on that material can be protected. In Thaler v. Perlmutter, the D.C. Circuit Court of Appeals applied the same principle: without identifiable human authorship, the result may fall outside copyright protection. Dominican companies operating with AI-generated assets without documenting human intervention may face registration objections before ONDA or evidentiary weaknesses in defending them as exclusive assets against third parties.

II. Ownership of Results: Internal Creation, Commissions, and Software

Once the first analysis is overcome (existence of a protectable work), the company needs a legal title allowing it to exploit the work. For works created in an employment relationship, Article 12 and what is agreed upon govern; for commissioned works (agencies, consultants), Article 14 requires examining the contract: what was commissioned, what rights were assigned, and for how long. In the absence of an express stipulation, the principal may not receive the expected usage rights.

The problem worsens with "vibe coding": applications, websites, chatbots, and automations generated through natural language instructions with AI assistance. Under Article 73 of Law 65-00, computer programs are protectable, but the presumption of assignment to the producer does not resolve who controls the asset when AI intervenes, nor the restrictions of tools, libraries, or incorporated APIs. The technology contract must define who controls the asset, what is delivered, and who is responsible if it fails. If the result includes distinctive signs, trade names, or branding elements, the trademark regime of Law 20-00 applies. The case Getty Images v. Stability AI [2025] EWHC 2863 (Ch) illustrates how AI-generated assets can trigger trademark and reputational risks independent of the copyright regime.

III. Confidential Information, Personal Data, and Liability for Content

Regardless of the authorship regime, uploading confidential information to external AI platforms may compromise trade secret protection (Article 178 of Law 20-00). When a company uploads contracts, code, or client data without controls, it weakens the argument of reasonable measures to preserve secrecy. If that data includes personal information, questions arise under Law 172-13 on personal data protection: informed consent, purpose, and transfer to external systems.

In 2023, engineers at Samsung Semiconductor uploaded proprietary source code and internal meeting transcripts to ChatGPT, leading the company to restrict the use of generative AI on its corporate networks. Cyberhaven reported that 2.3% of employees had entered confidential information into ChatGPT.

Moreover, liability for the content of generated documents (contracts with omitted clauses, inaccurate statements) falls under the law of obligations and rests on the person who signed and distributed the document, regardless of authorship. The U.S. Copyright Office also warned about voice and image replicas (deepfakes), which affect privacy and reputation; in the Dominican Republic, Article 52 of Law 65-00 offers protection against unauthorized commercial use of a person's bust or portrait, but synthesized voice and digital avatars lack specific regulation.

IV. Gaps in the Current Legal Framework

Law 65-00 does not regulate AI-assisted authorship, protection of digital replicas, use of protected works in model training, or minimum transparency in contracts involving generative tools. To date, the Dominican Republic lacks specific regulation on any of these points. In the European Union, Regulation (EU) 2024/1689 (EU AI Act) establishes transparency obligations, synthetic content labeling, and provisions on training models with protected works. Its scope as a comparative reference is relevant for Dominican companies with operations or exposure in European markets.

✅ Practical Considerations

Legal review of AI-generated assets cannot be limited to the final result. An image, document, app, or chatbot may function and still have an incomplete rights chain. A company or independent creator should not assume that every AI-generated result fully belongs to them. Before publishing, selling, or incorporating it into a brand, campaign, or software, they must check whether the asset can be legally protected and whether they have sufficient rights to use it.

           What is not reviewed beforehand will not be a preventive discussion: it will be with a client, a supplier, a platform, a competitor, an authority, or a court. In the age of AI, the fact that an asset works does not mean you can use it, sell it, or defend it legally.
        

About the author

Carlos Romero Polanco is a Dominican attorney and founder of LegalHub RD. He advises companies, institutions, and projects on public law, regulation, public procurement, strategic litigation, and intellectual property matters.

📚 For more interesting articles, visit www.legalhubrd.com