A technology company operating between Yerevan and a Western European hub deploys a generative AI system to produce marketing copy, software components, and visual assets. The output ships to clients across three jurisdictions. When a competitor reproduces that content without authorisation, the company's legal team faces an immediate and unsettling gap: Armenian intellectual property legislation was not written with machine-generated works in mind. Additionally. No court has yet resolved who. if anyone. owns the rights to output created autonomously by an AI system.
AI-generated works occupy a legal grey zone under Armenian intellectual property legislation, which conditions copyright protection on human creative authorship. No dedicated statutory provision addresses machine-generated content, and Armenian courts have not yet produced settled case law on ownership, authorship, or software liability for AI output. International businesses operating in Armenia or licensing AI-produced content to Armenian counterparties must therefore structure their rights defensively, relying on contractual arrangements and related-rights mechanisms until the legislative regime catches up.
This analysis examines the doctrinal foundations of Armenian copyright law as applied to AI output, identifies where competing interpretations diverge. Maps the gap between statute and practice. Additionally, draws out the cross-border implications for CIS-region clients. It closes with strategic recommendations and a forward-looking outlook on regulatory trajectory.
Doctrinal foundations: what Armenian IP law says – and what it does not
Armenian intellectual property legislation traces its architecture to a civil law tradition that centralises the human author as the originatory subject of copyright. The concept of hեղinakanutyun – or authorship – is defined by reference to creative effort exercised by a natural person. This is not a quirk of drafting. It reflects a deliberate doctrinal choice shared across much of the CIS legislative tradition: rights arise from human creative activity, not from any process, tool, or automated system.
Under Armenian copyright and related rights legislation, a work qualifies for protection if it is original and expressed in a perceivable form. Originality, in the civil law sense used in Armenia, requires that the work bear the imprint of the author's personality – a threshold that AI output cannot meet autonomously, because no personality is present. This is the first doctrinal barrier for AI-generated content.
A second barrier arises from the structure of moral rights. Armenian intellectual property legislation grants authors a bundle of personal rights – including the right of attribution and the right of integrity – that are inalienable and cannot be transferred. These rights attach to a human creator. They cannot, as a matter of doctrine, vest in a machine or in the operator of a machine by default. The operator must establish a separate legal basis for any claim.
Related rights – covering producers of phonograms, broadcasters, and database makers – offer a partial workaround. Under Armenian law, database protection can arise from substantial investment in the collection or arrangement of data, without requiring originality in the copyright sense. For AI systems that produce structured datasets or compilations, this branch of intellectual property legislation may offer protection where copyright cannot. Practitioners in Armenia increasingly note that the database right route is underused by technology companies entering the market.
What Armenian legislation does not contain is equally significant. There is no provision analogous to the United Kingdom's copyright rule for computer-generated works, which vests rights in the person who makes the necessary arrangements for the creation of the work. There is no sui generis category for AI-assisted output. There is no definition of "AI system" in the IP statute. This absence is not unique to Armenia. it mirrors conditions across most CIS jurisdictions. but it creates a specific set of risks for businesses that treat Armenian law as a passive backdrop to their technology licensing arrangements.
Competing interpretations and the gap between statute and practice
In the absence of dedicated legislation and settled court practice, at least three competing interpretive positions have emerged among Armenian practitioners and commentators.
The first position holds that AI-generated output is simply unprotected. If no human author contributed original creative expression, the work enters the public domain upon creation. Under this view, any person – including a competitor – may reproduce, adapt, or distribute AI-generated content without infringing Armenian copyright. This is the most textually defensible reading of the current statute. It is also the most commercially damaging for AI developers and users.
The second position argues for derivative human authorship. Where a human operator exercises meaningful creative direction over the AI system – selecting parameters, curating output, arranging elements – the final work may qualify as an authored work with the human as its creator. The AI is treated as a sophisticated tool, analogous to a camera or editing suite. This interpretation has practical appeal, but its application is fact-specific. Courts would need to assess, on a case-by-case basis, the degree of human creative contribution relative to the machine's autonomous generation. Armenian courts have not yet articulated a threshold for this analysis.
The third position rests on contractual and assignment mechanisms. Even if the work itself cannot carry copyright, the technology licensing agreement between the AI developer and the deploying business can allocate commercial control over the output. Confidentiality obligations, restrictions on reproduction, and database right protections can together replicate much of what copyright would otherwise provide. This position is the most practically viable in the short term, though it depends entirely on the quality of the underlying technology licensing documentation.
The gap between statute and practice manifests acutely in enforcement. Armenian civil procedure rules permit intellectual property right holders to seek injunctions and damages through the courts of general jurisdiction. But the claimant must establish standing – which requires demonstrating an enforceable right. A business relying on the derivative human authorship theory must be prepared to litigate the authorship question before it can even reach the infringement claim. This procedural exposure creates leverage for infringers and disincentivises enforcement. Specialists in the Armenian market note that many businesses absorb infringement losses rather than litigate an unresolved doctrinal question.
Algorithmic accountability presents a further dimension. Where an AI system generates content that itself infringes a third party's rights – reproducing protected expression from training data without authorisation – the question of software liability falls on the deploying business. Armenian tort law and intellectual property legislation together suggest that the operator, not the developer, bears primary exposure in the absence of a contractual indemnity. This is consistent with the general principle that the person who deploys a tool in commerce bears responsibility for its outputs. However, the precise contours of this software liability allocation have not been tested before Armenian courts.
For a deeper comparative perspective on how a neighbouring CIS jurisdiction has begun to address these questions. See our analysis of AI-generated works and intellectual property in Russia. This illustrates a parallel doctrinal trajectory with distinct legislative responses.
Cross-border implications for CIS clients and international businesses
Armenia's position within the CIS creates both opportunities and complications for international businesses structuring AI-related IP across the region. Armenia is party to the Berne Convention and the Agreement on Trade-Related Aspects of Intellectual Property Rights, which sets minimum protection standards. These international instruments, however, address human authorship and do not resolve the AI-specific questions that Armenian domestic law has yet to answer. The international regime provides a floor, not a ceiling, and that floor was designed for a different technological era.
For businesses operating between Armenia and the European Union, the asymmetry is particularly pronounced. The EU's approach to AI regulation – through what practitioners now refer to as the AI Act compliance regime – imposes transparency, documentation, and risk-classification obligations on AI systems deployed in the EU market. Armenian law contains no comparable AI Act compliance structure. A business that deploys an AI content-generation system in both markets must therefore maintain two distinct compliance postures: one oriented toward EU algorithmic accountability standards, another toward the more permissive but legally uncertain Armenian environment.
Technology licensing between CIS parties raises its own complications. A licensor in one CIS state may seek to license AI-generated content to an Armenian licensee, relying on the assumption that the licensor holds valid copyright in the source jurisdiction. If that copyright is not recognised in Armenia – because the work fails the human authorship test – the licence is licensing nothing. The Armenian licensee acquires no enforceable rights and bears full exposure if a third party challenges its use of the content. Due diligence on the ownership chain of AI-generated licensed content is therefore essential for any inbound technology licensing transaction in Armenia.
Digital services businesses face a related set of questions around the treatment of AI-assisted deliverables under Armenian commercial legislation. Where a digital services contract promises deliverables that are in whole or in part AI-generated, the contract should specify the basis on which the service provider claims to transfer rights. The scope of the database rights being conveyed. Additionally, the allocation of software liability for infringing outputs. Without these provisions, the counterparty may find it has paid for a deliverable to which no transferable right attaches under Armenian law.
Enforcement of Armenian IP judgments abroad – and of foreign IP judgments in Armenia – adds a further procedural layer. Armenian civil procedure rules permit recognition of foreign judgments through a bilateral treaty regime. However, the absence of a specific bilateral treaty with many EU member states means that enforcement often proceeds through general reciprocity principles, which are applied inconsistently. A business that secures an IP judgment in Armenia relating to AI-generated content may find enforcement abroad limited by the importing court's own doctrinal scepticism about AI authorship.
To discuss how Armenian intellectual property legislation applies to your AI deployment or technology licensing arrangements, reach out to us at info@ferrazwhitmore.com.
Strategic recommendations for businesses operating in Armenia
Given the doctrinal uncertainty and the absence of settled court practice, businesses in Armenia with AI-generated IP interests should adopt a defensive, multi-layered approach. The following principles reflect current best practice among specialists advising on Armenian IP matters.
Document human creative contribution at every stage. For AI-assisted works where a human authorship claim is plausible. Maintain detailed records of the creative decisions made by human operators: prompt engineering choices, output selection criteria, editorial interventions, and arrangement decisions. This documentation forms the evidentiary basis for a derivative authorship claim if ownership is later challenged before an Armenian court.
Rely on database rights for structured outputs. Where AI systems produce datasets, compilations, or systematically organised content, assert database right protection under Armenian intellectual property legislation. This requires demonstrating substantial investment in obtaining, verifying, or presenting the database content. The threshold is investment-based, not originality-based, which makes it more accessible for AI-generated structured outputs.
Use technology licensing agreements to replicate copyright protection. Draft technology licensing agreements to impose reproduction restrictions, confidentiality obligations, and permitted-use definitions on all parties with access to AI-generated content. Where copyright protection is unavailable, contractual control can serve a similar commercial function. Ensure these agreements are governed by Armenian law or contain a choice-of-law clause that has been tested for enforceability in Armenian courts.
Allocate software liability explicitly in service contracts. Any digital services contract involving AI-generated deliverables should contain an explicit software liability allocation. The service provider should warrant the originality of the output to the extent achievable and indemnify the client against third-party IP claims arising from infringing training data. The client should acknowledge that the service provider cannot guarantee AI-generated output is free from similarity to existing works.
Monitor AI Act compliance developments in trading partner jurisdictions. Armenian businesses contracting with EU counterparties will face increasing pressure to align with EU AI Act compliance standards as a condition of commercial relationships. Proactive engagement with algorithmic accountability documentation – even absent a domestic legal obligation – reduces friction in cross-border commercial relationships and positions the business favourably for future regulatory convergence.
Our AI and technology law practice in Armenia provides tailored advice on structuring AI-related IP rights, drafting technology licensing agreements, and managing cross-border compliance obligations across CIS and EU markets.
Regulatory outlook and what to monitor
Armenia's legislative trajectory on AI-generated IP is at an early stage. Several indicators suggest that change is coming, though the pace and direction remain uncertain.
Armenia has engaged with the Eurasian Economic Union's working groups on digital economy regulation. Any harmonisation of IP rules across EEU member states would create pressure to address AI-generated works at a supranational level. Businesses with multi-jurisdictional CIS exposure should monitor EEU regulatory output for signals of an emerging regional standard.
Domestically, the Armenian government has signalled interest in developing a digital economy legislative agenda. Draft provisions addressing digital services, platform liability, and data governance have been under consideration. Whether AI-generated IP will be addressed as a discrete subject within this agenda – or left to emerge through court practice – is not yet determined. Practitioners note that reform is more likely to arrive incrementally, through case law and soft-law guidance, than through comprehensive legislation in the near term.
The influence of EU regulatory developments on Armenian law should not be underestimated. Armenia's Partnership and Cooperation Agreement with the EU creates an ongoing channel for legislative approximation. As EU-level thinking on AI Act compliance, algorithmic accountability. Additionally, the treatment of AI-generated content matures. Armenian legislators will face practical pressure to align. particularly as Armenian technology companies seek access to EU markets and EU investment.
For businesses already operating in Armenia, the optimal posture is to structure AI-related IP rights now on the assumption that the legislative environment will tighten. Retroactive restructuring of ownership chains is possible but costly. Establishing clean contractual foundations and documentation practices during this pre-legislative window is a significantly lower-cost investment than corrective action after a court challenge or regulatory intervention.
The interaction between Armenian IP legislation and broader CIS norms also warrants attention. Several CIS states have begun to address AI-generated content through copyright reform or soft-law guidance. As those positions consolidate, Armenian courts may draw on them as persuasive authority in the absence of domestic precedent. Understanding the regional direction of travel is therefore part of managing IP risk in Armenia today.
To explore how these developments affect your technology licensing or AI deployment strategy in Armenia, schedule a consultation at info@ferrazwhitmore.com.
Self-assessment: when to seek specialist advice on AI-generated IP in Armenia
This approach is most relevant to your situation if one or more of the following conditions applies.
- Your business deploys AI systems that generate content, software, or data in Armenia or for Armenian counterparties.
- You license AI-generated content to or from CIS-based parties and have not verified the ownership chain under Armenian law.
- Your digital services contracts include AI-assisted deliverables without explicit software liability and IP ownership provisions.
- You are subject to EU AI Act compliance obligations and also operate in Armenia, creating a dual-compliance posture.
- You are considering enforcement action in Armenia for infringement of AI-generated works and have not assessed the standing question.
Before initiating any IP registration, enforcement, or licensing transaction involving AI-generated content in Armenia, verify the following.
- Is there documented evidence of human creative contribution to the works in question?
- Do your technology licensing agreements explicitly address AI-generated output and its ownership status under Armenian law?
- Have your digital services contracts been reviewed for software liability exposure under Armenian commercial and IP legislation?
- Have you assessed whether database right protection is available as a fallback for structured AI outputs?
- Is your cross-border enforcement strategy consistent with Armenian civil procedure rules on recognition of foreign judgments?
For a tailored strategy on AI-generated IP protection and technology licensing in Armenia, reach out to our team at info@ferrazwhitmore.com.
Frequently asked questions
Q: Can an AI system hold copyright in Armenia?
A: No. Armenian intellectual property legislation ties copyright to human authorship. An AI system cannot hold rights in its own name. The practical question is whether the human who operated, trained. Alternatively. Directed the AI can claim authorship. and this depends on the degree of creative input they exercised. This Armenian courts have not yet resolved in consistent terms.
Q: How long does it take to register a copyright or related right in Armenia?
A: Copyright in Armenia arises automatically upon creation. Voluntary registration with the relevant state authority typically takes several weeks, depending on documentation and the type of work. Registration is not required for the right to exist, but it creates a rebuttable presumption of authorship and is strongly advisable for AI-assisted works where ownership may later be disputed.
Q: Is it a misconception that Armenian law simply mirrors Russian IP rules on AI?
A: Yes. While Armenian intellectual property legislation shares historical roots with the broader CIS legislative tradition, Armenia has developed its own copyright and related rights regime. Key differences exist in scope, enforcement mechanisms, and the treatment of digital services. Engaging a lawyer in Armenia with dedicated IP expertise is essential before assuming that approaches valid in other CIS states will apply.
About Ferraz & Whitmore
Ferraz & Whitmore is an international law firm based in Lisbon, advising business clients across 46 jurisdictions. Our team combines Portuguese civil law expertise with English common law tradition to deliver cross-border legal solutions in AI and technology law, intellectual property protection, and technology licensing across CIS and European markets. We advise technology companies, digital services providers, and institutional investors operating in Armenia and the wider CIS region on IP ownership structuring, software liability allocation, and AI Act compliance positioning. Our attorneys have experience advising on cross-border IP and AI matters before arbitral bodies and courts across both civil law and common law systems. The firm's Lisbon base provides direct access to EU regulatory developments, enabling us to align our CIS-market advice with the direction of EU algorithmic accountability standards. As an international law firm in Armenia and across the CIS, Ferraz & Whitmore supports clients who need results-oriented counsel at the intersection of technology and law. For advice on intellectual property matters in Armenia, contact us at info@ferrazwhitmore.com.
Disclaimer: This publication is provided for informational purposes only and does not constitute legal advice. The information herein should not be relied upon as a substitute for professional legal counsel tailored to your specific circumstances. Ferraz & Whitmore assumes no liability for actions taken or not taken based on the contents of this material. For advice regarding your particular situation, please contact info@ferrazwhitmore.com.