A technology company launches a marketing campaign in Hungary. Every visual asset, every tagline, and every background score is produced entirely by a generative AI platform. The campaign succeeds commercially. Then a competitor claims that one AI-generated image infringes its registered design rights. The technology company's legal team faces an immediate question: who owns the infringing work, and who bears responsibility for it? Hungarian intellectual property legislation was not designed with this scenario in mind. Yet these questions are now reaching Hungarian courts and the Hungarian Intellectual Property Office with increasing frequency.
AI-generated works in Hungary occupy a legal grey zone under existing intellectual property legislation, which conditions copyright protection on human authorship and creative choice. The gap between statute and emerging commercial practice is widening as generative AI tools proliferate across Hungarian and European markets. Businesses operating in Hungary must now assess ownership, liability, and AI Act compliance obligations before deploying AI-generated content commercially.
This analysis examines the doctrinal foundations of Hungarian copyright and intellectual property law as they apply to AI-generated outputs, maps the emerging tensions in court and administrative practice. Addresses cross-border implications for European clients. Additionally, sets out a strategic approach for businesses that generate or exploit AI-produced works in Hungary.
Doctrinal foundations: what Hungarian intellectual property law says about authorship
Hungarian intellectual property legislation follows the European civil law tradition in grounding copyright protection in the concept of individual creative expression. A work qualifies for protection when it reflects the personal intellectual creation of a natural person. This requirement is not merely procedural. It is the structural foundation on which ownership, duration, and transferability of rights are built.
An AI system, regardless of its sophistication, is not a natural person under Hungarian law. It cannot hold rights, assume obligations, or be identified as an author. This position is consistent with the approach adopted across the EU and aligns with the broader civil law understanding of legal personality. The consequence is direct: a work produced autonomously by an AI system – without meaningful human creative input – does not qualify for copyright protection under Hungarian intellectual property legislation.
The difficulty begins when the analysis moves beyond the purely autonomous case. In practice, most commercially deployed AI systems operate through a layered human-AI interaction. A human operator selects the model, drafts the prompt, adjusts parameters, curates outputs, and applies the result to a specific commercial purpose. At each stage, some degree of human judgment is exercised. The question Hungarian law must answer – and has not yet answered definitively – is how much human creative contribution is required to satisfy the authorship threshold.
Practitioners in Hungary note that courts have historically applied the authorship threshold with considerable flexibility in analogous contexts, such as software development and computer-assisted design. The dominant approach has been to look for evidence of individual creative choices that distinguish the work from what any competent operator would produce mechanically. Applied to AI-generated works, this logic suggests that a human who makes substantive creative decisions during the generation process. selecting distinctive aesthetic parameters. Iterating on outputs with a clear artistic intention. Alternatively, substantially editing the raw AI output. may satisfy the authorship requirement.
However, where a human's role is limited to issuing a generic prompt and accepting the first output, the creative contribution may fall below the required threshold. In that scenario, the work enters a contested space: it is not clearly protected under Hungarian copyright legislation, yet it may still carry commercial value. Businesses that rely on such outputs without understanding this gap risk discovering that their most valuable creative assets are unprotected – and therefore freely reproducible by competitors.
Ownership disputes and the gap between statute and practice
Even where a human creative contribution is sufficient to support copyright protection, a secondary question arises immediately: who owns the resulting rights? Hungarian intellectual property legislation provides default rules for works created in employment and works produced under commission contracts. These rules were drafted for human authors. Their application to AI-assisted workflows is not always straightforward.
Under employment-related provisions of Hungarian copyright legislation, rights in works created by an employee in the course of employment vest automatically in the employer, subject to exceptions. This default rule is broadly understood and widely applied in the software development context. Its extension to AI-generated content is less settled. Where an employee uses a company-licensed AI platform to produce content as part of their assigned duties, the employer's claim to ownership is defensible. Where the same employee uses a personal AI subscription for a work-related task, the position is less clear.
Commission arrangements introduce further complexity. Hungarian intellectual property legislation requires that transfers and licences of copyright be made in writing and specify the rights transferred with sufficient precision. A commissioning contract that pre-dates the widespread use of generative AI will almost certainly fail to address AI-generated outputs explicitly. This gap is not merely theoretical. Disputes between commissioners and commissioned parties over ownership of AI-generated deliverables are already arising in Hungarian commercial practice.
Technology licensing adds another layer. When a business uses a commercial AI platform to generate content. The platform's terms of service define. at least contractually. how intellectual property rights in the output are allocated between the platform provider and the user. These terms vary substantially across providers. Some assert broad licences over outputs; others purport to assign all rights to the user. None of these contractual arrangements alter the underlying position under Hungarian intellectual property legislation: contractual ownership claims cannot create copyright protection where the statutory threshold for authorship is not met.
The Fővárosi Törvényszék (Metropolitan Court of Budapest) serves as the specialist first-instance court for intellectual property disputes in Hungary. It has handled disputes involving computer-generated works in adjacent fields, including software-generated music and automated financial reports. The court's emerging approach focuses on the degree of human editorial control exercised over the final output. Works where a human made identifiable choices that shaped the distinctive character of the output have been treated more favourably than works where the human role was purely operational.
For businesses operating as a lawyer in Hungary or relying on Hungarian counsel, the practical lesson is that ownership of AI-generated works should be documented proactively. Contemporaneous records of the prompting process, parameter choices. Additionally. Editorial decisions taken by human operators provide the evidentiary foundation needed to defend an ownership claim before the Metropolitan Court or the Magyar Szabadalmi Hivatal (Hungarian Intellectual Property Office).
To explore how AI-generated content ownership questions interact with broader intellectual property strategy in Hungary, see our practice overview of intellectual property law in Hungary.
Software liability, algorithmic accountability, and the EU AI Act
The ownership question is only one dimension of the emerging legal challenge. Equally pressing for Hungarian businesses is the question of liability when AI-generated works cause harm. whether by infringing third-party rights. By producing defamatory content. Alternatively, by generating outputs that are misleading in a regulated commercial context.
Hungarian civil liability rules follow the general framework of civil law systems: liability attaches to legal persons or natural persons who cause damage through fault or, in specific cases, regardless of fault. An AI system is neither. This creates an attribution gap that Hungarian courts must fill by reference to existing principles. The most relevant are software liability and product liability doctrines, both of which have application to AI systems deployed as digital services.
Under Hungarian civil legislation and the EU product liability regime as applied in Hungary, the developer or deployer of a defective system may be liable for damage caused by that system's outputs. The application to AI is developing but the direction is clear: liability will attach to the entity that placed the AI system into commerce or that deployed it in a specific operational context. This is consistent with the principle of algorithmic accountability – the idea that someone in the supply chain must be answerable for the outputs of automated systems.
The EU AI Act, which applies directly across all member states including Hungary, introduces a structured risk-classification system for AI systems. Systems used to generate content – including text, images, audio, and video – are subject to transparency obligations. Providers must ensure that AI-generated content is labelled as such where there is a realistic risk of confusion with human-produced content. This transparency requirement has immediate practical significance for Hungarian businesses in media, advertising, and digital services.
AI Act compliance for Hungarian businesses involves several specific steps. First, the system must be classified according to the Act's risk tiers. General-purpose AI models used for content generation are subject to the Act's general-purpose AI provisions, which impose documentation, transparency, and copyright compliance obligations on providers. Second, deployers – Hungarian businesses that use a general-purpose model for a specific application – bear responsibility for ensuring that the deployment context does not create high-risk conditions not addressed by the provider. Third, businesses must implement disclosure mechanisms that satisfy the Act's transparency requirements in their specific commercial context.
A common misconception among Hungarian technology businesses is that AI Act compliance is the provider's problem alone. In practice, the Act imposes a shared responsibility model. A Hungarian media company that deploys an AI content-generation tool and publishes the outputs without disclosure is exposed to enforcement action under the Act, regardless of whether the underlying model's provider is also in breach. Enforcement in Hungary falls within the remit of the designated national supervisory authority, which has powers to impose administrative sanctions.
For a detailed assessment of AI technology regulation obligations applicable to your business in Hungary, contact us at info@ferrazwhitmore.com.
Cross-border implications for European clients
For international clients – whether a German media group licensing AI-generated content to a Hungarian subsidiary, a Dutch technology company offering a generative AI platform to Hungarian customers. Alternatively. A British brand using AI-produced visuals in pan-European campaigns that include Hungary – the cross-border dimension of this analysis is often the most commercially significant.
Hungary is a member of the EU and party to the principal international intellectual property conventions. The Berni Egyezmény (Berne Convention) provides the baseline for cross-border copyright recognition. Under its national treatment principle, a work protected by copyright in one contracting state must be accorded equivalent protection in other contracting states. The difficulty for AI-generated works is that this protection-by-reciprocity mechanism presupposes that the work qualifies for protection in at least one jurisdiction. Where the work fails the authorship threshold in Hungary and in the state of origin, there is no protection to recognise.
This creates a specific risk for European clients who assume that AI-generated works protected under more permissive interpretations in their home jurisdiction will carry equivalent protection into Hungary. The United Kingdom, for example, has a specific statutory provision in its copyright legislation that extends protection to computer-generated works without a human author. Attributing ownership to the person who made the necessary arrangements for the work's creation. A UK business holding copyright in such a work under English law cannot automatically assert equivalent protection before Hungarian courts. Hungarian judges will apply Hungarian authorship standards to assess whether protection exists locally.
The interaction between Hungarian copyright legislation and EU harmonisation measures is also relevant. EU directives on copyright in the digital single market introduce text and data mining exceptions and impose obligations on online platforms regarding user-uploaded content. These measures shape the environment in which AI training datasets are assembled and AI outputs are distributed. A Hungarian AI developer training a model on web-scraped content must comply with both the statutory exceptions and any contractual restrictions applicable to the data sources. and must do so under Hungarian and EU law simultaneously.
Technology licensing between EU jurisdictions raises further questions. Where a German parent company licenses an AI platform to its Hungarian subsidiary for internal content generation. The licence must address intellectual property ownership of outputs, liability allocation for infringing outputs. Additionally, compliance obligations under the EU AI Act. The licence terms that work effectively in a purely German context may not translate directly into the Hungarian legal environment, particularly regarding the formal requirements for copyright transfer under Hungarian intellectual property legislation.
Practitioners advising European clients on Hungary-facing AI strategies recommend a jurisdiction-specific review of three elements: first, whether the AI-generated works at issue meet Hungary's authorship threshold. second. Whether the contractual arrangements governing ownership and liability are effective under Hungarian law. and third, whether the deployment model satisfies AI Act compliance requirements as applied by Hungarian regulatory authorities. Skipping this review – particularly when a client is scaling a successful AI content strategy from one EU market to another – is one of the most common sources of avoidable legal exposure.
For a comparative perspective on how similar questions are addressed in another EU civil law jurisdiction, our analysis of AI-generated works and intellectual property in Portugal provides a useful parallel.
Strategic recommendations and the competitive dimension
The legal uncertainty surrounding AI-generated works in Hungary is not symmetrical in its commercial effects. Businesses that address it proactively gain a defensible competitive position. Those that do not face the risk of discovering, at the worst possible moment, that their AI-generated assets are unprotected, their liability exposure is unquantified, and their commercial arrangements are unenforceable.
The first strategic priority is documentation. Businesses should establish internal protocols that capture the human creative contribution at each stage of the AI content generation process. This means recording prompt strategies, parameter choices, selection criteria, and editorial revisions. The record does not need to be elaborate. It needs to be contemporaneous and specific enough to demonstrate that identifiable human choices shaped the final output. This documentation serves a dual purpose: it supports an ownership claim under Hungarian copyright legislation, and it provides evidence of the human oversight required by the EU AI Act's transparency and accountability provisions.
The second priority is contractual clarity. Every commercial arrangement that involves AI-generated content. whether a commission contract, an employment agreement, a technology licence. Alternatively. A distribution arrangement. should address ownership of AI-generated outputs, liability allocation for infringing or harmful outputs. Additionally, compliance responsibilities under applicable AI regulation. Generic intellectual property clauses drafted before the generative AI era will not achieve this. Businesses should review and update their standard contracts to reflect the specific conditions under which AI tools are used.
The third priority is risk classification under the EU AI Act. Not all AI content generation activities carry the same regulatory exposure. A business that uses AI to generate internal reports faces different obligations than one that deploys AI to produce consumer-facing advertising. The Act's risk classification logic requires a fact-specific assessment. Businesses that complete this assessment early are better positioned to design compliant workflows and to respond to regulatory inquiries without disruption to commercial operations.
The fourth element is enforcement strategy. Where a business holds protectable AI-assisted works in Hungary, enforcement through the Hungarian Intellectual Property Office and the Metropolitan Court of Budapest requires a litigation-ready evidence package. This means clear ownership documentation, evidence of the human creative contribution, and – where relevant – a clear record of the infringing party's access to and use of the protected work. Hungarian procedural rules in intellectual property matters require that evidence be assembled and disclosed at an early stage of proceedings. Businesses that prepare this material proactively are in a substantially stronger position than those that attempt to reconstruct it after a dispute arises.
The competitive dimension of this analysis is worth emphasising for businesses currently evaluating whether to invest in AI-generated content at scale. The businesses that will extract the most commercial value from AI-generated works in Hungary are those that structure their AI workflows to produce protectable outputs. Document their creative contribution with sufficient rigour. Additionally, maintain the contractual and regulatory infrastructure needed to defend and enforce their rights. Those that treat AI content generation as a purely operational matter – and defer the legal structuring – are, in effect, subsidising their competitors' freedom to copy.
To receive a tailored assessment of your AI content strategy and intellectual property position in Hungary, reach out to info@ferrazwhitmore.com.
Outlook: where Hungarian law is heading
Hungarian intellectual property legislation is not static. The pressure to address AI-generated works directly is building from multiple directions simultaneously. At the EU level, the AI Act creates a regulatory environment that implicitly presupposes that AI-generated content is identifiable, attributable, and subject to legal oversight. The EU's ongoing work on copyright harmonisation is likely to address the authorship gap more explicitly in the coming years. Hungary, as an EU member state, will be required to implement any resulting directives.
At the national level, the Hungarian Intellectual Property Office has signalled awareness of the gap between existing copyright doctrine and the realities of AI-generated content. While formal legislative proposals have not yet reached the parliamentary stage, the administrative practice of the Office is evolving. Practitioners in Hungary observe a growing willingness to engage with AI-specific arguments in examination and opposition proceedings, even where the applicable statutory provisions do not yet explicitly address AI.
The courts are moving more cautiously. The Metropolitan Court of Budapest has not yet issued a decision that directly and comprehensively addresses the ownership of AI-generated works. The cases that have come before it have tended to resolve on narrower grounds – for example, whether a specific level of human editorial involvement was present – without establishing a broad doctrinal framework. This case-by-case approach is consistent with the civil law tradition but leaves businesses without the clear precedent that would support confident commercial planning.
The most likely near-term development is legislative intervention at the EU level, followed by transposition into Hungarian law. International clients should monitor this process closely. The direction of travel – toward explicit recognition of some form of protection for AI-assisted works, combined with stronger transparency and accountability requirements for deployers – is reasonably clear. The precise mechanism and the threshold for protection are not. Businesses that position themselves now for a range of regulatory outcomes are better placed to adapt quickly when the legislative picture clarifies.
For businesses currently scaling AI content strategies across multiple EU jurisdictions, Hungary presents both a challenge and an opportunity. The challenge is the current uncertainty. The opportunity is that businesses that invest in legally sound AI content workflows in Hungary today will be structurally advantaged when the regulatory environment firms up. and when competitors who deferred the legal work find themselves scrambling to catch up.
Our AI and technology law practice in Hungary advises on the full spectrum of issues covered in this analysis, from ownership structuring and contract drafting to AI Act compliance and enforcement strategy.
Frequently asked questions
Q: Can an AI system be named as the author of a copyrighted work in Hungary?
A: No. Hungarian copyright legislation reserves authorship for natural persons. An AI system cannot hold copyright. Ownership typically falls to the person or entity that directed the creative process – most commonly the developer or the commissioning business – but this depends on the degree of human contribution involved.
Q: How does the EU AI Act affect Hungarian businesses that use AI to generate content?
A: The EU AI Act introduces tiered obligations based on risk classification. Hungarian businesses deploying AI in content generation must assess whether their systems fall into high-risk categories, implement transparency disclosures, and maintain technical documentation. AI Act compliance requirements apply directly as EU law, without a separate transposition step.
Q: What is the typical timeline and cost for resolving an AI-related IP dispute in Hungary?
A: Administrative proceedings before the Hungarian Intellectual Property Office typically conclude within several months. Court litigation before the Metropolitan Court of Budapest – the specialist IP venue – can extend to one to three years at first instance, depending on complexity. Legal fees in Hungary start from several thousand euros for advisory work and rise significantly for contested proceedings. Engaging a lawyer in Hungary with specialist AI and intellectual property experience at an early stage typically reduces the overall cost and duration of proceedings.
About Ferraz & Whitmore
Ferraz & Whitmore is an international law firm based in Lisbon, advising business clients across 46 jurisdictions. Our AI and technology law practice covers the full lifecycle of AI-related intellectual property matters in Hungary and across European markets. from ownership structuring and technology licensing to AI Act compliance and enforcement proceedings before the Hungarian Intellectual Property Office. We combine Portuguese civil law expertise with English common law tradition, giving us a distinctive perspective on how divergent legal systems handle the same AI-generated works and software liability questions. As a law firm in Hungary-facing matters, we work with international technology companies, media groups, institutional investors, and in-house legal teams who need cross-border legal solutions grounded in the specific conditions of each target jurisdiction. Our attorneys have advised on digital services deployments, algorithmic accountability frameworks, and technology licensing arrangements across civil law and common law systems. To discuss your AI intellectual property strategy in Hungary, 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.