A technology company deploying a generative AI system in Minsk discovers that its competitor has copied the AI's output verbatim – imagery, code, and marketing text. The company's lawyers search for a clear enforcement path. They find instead a body of intellectual property legislation drafted before generative AI existed, a registry that has never processed an AI-authorship claim, and a regional digital-services environment undergoing rapid but uncoordinated change. The gap between commercial reality and legal protection is wide.
AI-generated works in Belarus currently occupy a doctrinal grey zone under existing intellectual property legislation, which conditions copyright protection on human creative authorship. No dedicated statute addresses algorithmic accountability or software liability for AI-produced content. Businesses operating in or through Belarus must therefore structure their IP positions around available proximate instruments – technology licensing agreements, database rights, and trade secret protection – while monitoring legislative developments closely.
This analysis examines the doctrinal foundation, the gap between statute and practice, competing interpretive approaches, cross-border implications for CIS clients, and the strategic options available to international businesses right now.
Doctrinal background: what Belarusian intellectual property law actually says
Belarusian intellectual property legislation traces its conceptual lineage to Soviet-era copyright doctrine. That doctrine placed human creative effort at the centre of every protected work. The post-Soviet codification preserved this foundation. Copyright in Belarus attaches to an author – a natural person whose intellectual creative activity produces the work.
This baseline rule creates an immediate problem for AI-generated outputs. A generative model is not a natural person. It does not have legal personality under Belarusian civil legislation. It cannot hold rights, assert claims, or transfer entitlements. Outputs it produces autonomously therefore do not fit neatly into any protected category.
The position is not unique to Belarus. Jurisdictions across the CIS region share this structural gap. Practitioners in the region note that the issue was largely theoretical until the commercial deployment of large language models and generative image tools accelerated sharply. What was once an academic question is now a daily operational issue for technology companies, media businesses, and software developers.
Belarusian intellectual property legislation does recognise several rights that do not depend on human authorship in the same way copyright does. Database rights protect the substantial investment made in compiling, verifying, and presenting data. Computer program protection – treated similarly to literary works – requires human authorial contribution at the programming level. However. That contribution may be found in the creation of the AI system itself rather than in each individual output.
Trade secret protection under Belarusian commercial legislation offers a third avenue. An AI-generated output can qualify as a trade secret if it meets the conditions of commercial value, confidentiality, and active protection measures. This approach sidesteps the authorship question entirely. Its limitation is obvious: trade secret protection evaporates the moment the information enters the public domain.
Technology licensing under Belarusian civil and commercial legislation provides a fourth instrument. Businesses can define by contract what rights attach to AI-generated outputs within a licensed system. The enforceability of such clauses has not been tested extensively before Belarusian courts, but the general principle of freedom of contract supports their validity. The Natsionalny tsentr intellektualnoy sobstvennosti (National Centre for Intellectual Property) operates the voluntary registration system for works and related rights. Registration does not create copyright, but it establishes a rebuttable presumption of authorship and a fixed priority date – both valuable in any subsequent dispute.
For businesses engaged in digital services in Belarus, the practical takeaway at the doctrinal level is this: the absence of explicit legislation is not a safe harbour. It is a vulnerability. An unprotected AI output can be copied, adapted, and commercialised by a competitor without triggering any clear legal remedy.
The gap between statute and practice: where courts and regulators stand
Belarusian courts have not yet produced a substantial body of case law specifically addressing AI-generated works. The Verkhovny Sud (Supreme Court of Belarus) and the economic courts have handled IP disputes involving software and databases, but generative AI authorship has not been squarely adjudicated as a distinct issue in published decisions.
This absence of precedent creates interpretive uncertainty in both directions. On one reading, courts may apply the existing human-authorship requirement strictly. Under this approach, an AI-generated output without a traceable human creative contribution receives no copyright protection at all. The output enters a form of public domain. Third parties may use it freely.
On a competing reading, courts could identify the human author in the person who designed the AI system, curated the training data, or crafted the prompts that generated the specific output. This approach stretches the existing doctrinal framework rather than breaking it. Practitioners in CIS jurisdictions note that courts have historically favoured purposive interpretation when faced with novel technology questions.
A third interpretive possibility draws on the concept of sotvorcheskaya deyatelnost (co-creative activity) recognised in some civil law systems. Under this reading, the human operator of an AI system makes a creative contribution through their choices – selecting the system, configuring parameters, directing outputs. The resulting work is treated as a human-authored work with AI assistance, not an AI-produced work without human authorship. This interpretation is the most commercially useful for businesses. It is also the most legally contested.
The Ministerstvo ekonomiki (Ministry of Economy of Belarus) and the relevant digital-economy regulatory bodies have issued general guidance on the Hi-Tech Park (High-Technology Park) regime – Belarus's principal vehicle for technology business incentives. That guidance addresses software development and digital services broadly. It does not resolve the authorship question for AI-generated content specifically. Algorithmic accountability for AI outputs remains an undeveloped area at the regulatory level.
The practical consequence is that a business seeking to enforce rights in an AI-generated work before a Belarusian economic court today would face genuine doctrinal uncertainty at the threshold. The outcome would depend substantially on how the specific judge approaches the gap in the legislation – and on how thoroughly the claimant has documented the human creative contribution made during the AI-assisted production process.
For international clients, this uncertainty has a concrete cost. A competitor who copies an AI-generated product catalogue, dataset, or piece of software can advance a credible argument that no copyright subsists. The burden shifts to the originator to establish a protected interest. Without advance documentation, that burden may be impossible to discharge.
Cross-border implications for CIS clients and the regional regulatory trajectory
Belarus is a member of the Evraziysky ekonomichesky soyuz (Eurasian Economic Union, EEU). The EEU has developed harmonised intellectual property rules across its member states, which include Russia, Kazakhstan, Armenia, and Kyrgyzstan alongside Belarus. Those harmonised rules share the same foundational gap: they were designed for human authors and have not been updated to address AI authorship.
For a business operating across CIS markets, the structural problem therefore compounds. An AI-generated output produced in Belarus and commercialised in Russia or Kazakhstan faces the same doctrinal uncertainty in each jurisdiction. Our parallel analysis of AI-generated works and IP questions in Russia explores how Russian courts and regulators are beginning to address this challenge. and the divergences between Russian and Belarusian approaches are already becoming commercially significant.
The EU's approach to AI regulation – including AI Act compliance obligations and algorithmic accountability requirements – does not apply directly within Belarus. However, it shapes the environment for Belarusian technology companies that export digital services into EU markets or that partner with EU-based businesses. A Belarusian software developer supplying an AI tool to a German or Polish client must contend with the EU regulatory regime at the delivery end of the transaction. Even if Belarusian domestic law is silent on the point.
Technology licensing structures designed for cross-border CIS transactions should therefore address both ends of the chain. The agreement should define what rights attach to AI-generated outputs under the applicable law of each jurisdiction. Specify which party bears the risk of unprotectability. Additionally, include representations about the human creative contribution embedded in the AI system. These provisions are increasingly standard in well-drafted technology licensing agreements for regional deployments.
Software liability for AI outputs is a related pressure point. Where an AI-generated work causes harm – a defamatory text, an infringing image, an erroneous medical recommendation – the question of who bears liability under Belarusian civil legislation is unresolved. The developer of the AI system, the operator who deployed it, and the entity that used the output all have potential exposure. Allocating that exposure contractually is possible; doing so without specialised advice creates substantial residual risk.
CIS investors and technology entrepreneurs who have structured Belarusian IP holdings as part of a regional portfolio should conduct a specific review of how their AI-generated assets are characterised. Works registered voluntarily with the National Centre for Intellectual Property before any dispute arises are in a materially stronger position than unregistered outputs. The registration process is accessible and the costs are modest relative to the protection it provides.
For a tailored strategy on intellectual property structuring and AI-generated content protection in Belarus, reach out to info@ferrazwhitmore.com.
Strategic options and the Ferraz & Whitmore perspective
The absence of settled law in Belarus does not mean businesses must operate without protection. It means protection must be engineered deliberately rather than assumed automatically.
The most effective approach layers multiple instruments. Copyright – where a human creative contribution can be documented – provides the broadest protection and the clearest enforcement pathway. Technology licensing defines rights by contract where copyright is uncertain. Database rights cover compilations of AI-generated content where the investment in assembly is substantial. Trade secret protection covers commercially valuable outputs that can be kept confidential. Each layer addresses a different vulnerability; together they create a defensible position.
Businesses engaged in AI-assisted content creation should implement documentation practices from the outset. This means recording the human choices made at each stage of the production process: system selection, training data curation, prompt engineering, output selection, and post-generation editing. The more granular the record, the stronger the argument for human authorship. Courts in civil law systems – and Belarusian courts specifically – respond to documentary evidence. An absence of documentation is treated as an absence of the fact itself.
The AI and technology law practice at Ferraz & Whitmore in Belarus works with technology companies, regional investors, and in-house legal teams to structure these documentation and registration programmes. The work is preventive. Once a competitor has copied an output and litigation is underway, the options narrow considerably.
For businesses that already hold AI-generated assets without adequate documentation, a retrospective audit is the first priority. The audit identifies which outputs have a defensible human-authorship argument, which are better protected under alternative instruments, and which may need to be reproduced with proper documentation before they can be commercialised safely.
Contractual structures also require attention in the context of employment legislation. Where AI-generated works are produced by employees using employer-provided systems, Belarusian employment legislation and IP legislation interact to determine who owns the output. The default position under Belarusian law assigns rights in works created in the course of employment to the employer. but that default is conditioned on the work qualifying as a protected work in the first place. If the AI-generated output lacks copyright protection, the employment-law assignment rule does not engage. The employer may hold no rights at all unless the employment contract or a separate IP agreement addresses the gap explicitly.
This interaction between employment legislation and intellectual property legislation is one of the least-understood risk areas for international companies deploying AI tools in Belarus. A practical example illustrates the point. A foreign company's Minsk-based development team uses a generative coding tool to produce software components. The components are AI-generated with minimal human editing. The employment contracts are standard. No separate IP agreement assigns AI outputs to the company. The result is a genuine ownership uncertainty over the software – uncertainty that surfaces only when the company seeks to enforce rights against a third party or sell the asset.
The intellectual property practice in Belarus at Ferraz & Whitmore advises on the full spectrum of these structuring questions, from registration strategy to contractual allocation of AI-generated content rights.
Looking at the regulatory outlook, Belarus has signalled interest in developing a more comprehensive digital-economy legislative regime. The High-Technology Park framework – which already provides a distinct regulatory environment for technology companies – is the most likely vehicle for any future AI-specific rules. Businesses with a presence in the HTP should monitor legislative consultations closely. Early engagement in regulatory dialogue on AI and IP often shapes the final legislative text in civil law jurisdictions where regulatory drafting is concentrated in a small number of government bodies.
At the regional level, the EEU's intellectual property harmonisation process creates the possibility of a coordinated CIS response to the AI authorship question. That process moves slowly. Businesses should not rely on it for near-term protection. But the direction of travel. towards recognising some form of protected status for AI-assisted works, with human contribution as the qualifying criterion – is consistent across the jurisdictions that have begun to address the question.
To explore legal options for protecting AI-generated works and structuring IP assets in Belarus, schedule a consultation at info@ferrazwhitmore.com.
Self-assessment: when to act and what to verify
A structured approach to AI and IP in Belarus is most urgently needed if any of the following conditions apply to your business.
- Your business produces or commercialises AI-generated content in Belarus or for Belarusian clients.
- Your AI-generated outputs have not been registered with the National Centre for Intellectual Property.
- Your technology licensing agreements do not define rights in AI-generated outputs explicitly.
- Your employment contracts predate the deployment of AI tools and do not address AI-generated works.
- You distribute AI-assisted digital services from Belarus into EU markets where AI Act compliance applies.
Before taking enforcement action or structuring a new AI-related transaction in Belarus, verify the following:
- Is there a documented record of the human creative contribution in each AI-generated work you rely on?
- Have all commercially significant AI outputs been submitted for voluntary registration?
- Do your technology licensing and employment agreements allocate AI output rights explicitly?
- Is your trade secret protection programme adequate for outputs that cannot qualify for copyright?
- Have you reviewed cross-border exposure in other CIS jurisdictions where the same outputs are used?
Frequently asked questions
Q: Can a company claim copyright over an AI-generated work under Belarusian law?
A: Under current Belarusian intellectual property legislation, copyright attaches to natural persons who make a creative contribution. A company cannot assert copyright solely by virtue of deploying an AI system. To build a defensible ownership position, businesses should document the human creative choices made during the AI-assisted production process – and retain that documentation carefully.
Q: How long does it take to register an AI-assisted work with the National Centre for Intellectual Property in Belarus?
A: Voluntary registration with the National Centre for Intellectual Property typically takes several weeks once the application package is complete. Registration is not mandatory for copyright subsistence in Belarus, but it provides important evidentiary weight in any subsequent dispute over authorship or priority date.
Q: Is it a misconception that AI-generated outputs automatically receive protection in Belarus?
A: Yes. Many businesses assume that outputs produced by a generative AI system are automatically protected as original works. Belarusian intellectual property legislation does not extend copyright to outputs that lack a human creative contribution. Works produced entirely by an AI system, without meaningful human authorial input. Exist in a legal grey zone and may be freely used by third parties unless a different legal basis. such as a database right or trade secret protection – can be established.
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 digital services regulation. In Belarus and across the CIS region, we advise technology companies, regional investors, and in-house legal teams on AI-generated content ownership, software liability, technology licensing, and algorithmic accountability. The firm's AI and technology law practice covers both EU-regulated markets and high-growth jurisdictions where regulatory regimes are still forming – including the full EEU territory. Our attorneys have advised on AI-related IP structuring across civil law and common law systems, and the firm participates in cross-border practice groups focused on emerging technology regulation. As an international law firm in Belarus and across CIS markets, Ferraz & Whitmore provides the specialist counsel that technology businesses need when statute has not yet caught up with practice. To discuss your situation, 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.