A technology company based in Tel Aviv deploys a generative AI system to produce marketing content, software code, and product imagery at scale. The output is commercially valuable. Yet when a competitor reproduces that content without permission, the company's lawyers discover an uncomfortable gap: Israeli intellectual property legislation was not designed with AI authorship in mind. Additionally. No court has yet authoritatively resolved who owns what an autonomous system creates.
AI-generated works occupy an unresolved space under Israeli intellectual property law. Current legislation predicates copyright on human authorship, leaving works produced autonomously by AI systems without a clearly designated owner. International businesses operating in Israel – or licensing AI-generated content into and out of the country – face real commercial exposure until the legislature or courts supply a definitive answer.
This analysis examines the doctrinal foundation of the problem, surveys competing interpretations emerging in Israeli legal practice, maps the gap between statute and day-to-day commercial reality. Addresses cross-border implications for clients active across the Middle East and Asia-Pacific region. Additionally, sets out the strategic steps that forward-looking businesses should consider now.
Doctrinal foundations: what Israeli intellectual property law says about authorship
Israeli intellectual property legislation draws its core concepts from a combination of common law heritage and civil law influence. That dual origin matters. Copyright protection under Israeli law attaches to an "original" work that reflects the personal creative effort of a human author. The requirement of human creative expression is not explicitly stated in every provision of the legislation. However. It is embedded in the doctrinal architecture through decades of judicial interpretation by the Beit Mishpat HaElyon (Supreme Court of Israel) and the district courts.
The originality threshold in Israeli copyright doctrine sits at a modest level. Courts have generally required only that a work reflect some degree of independent intellectual effort, rather than demanding a high standard of artistic merit. This low threshold was set with human creativity in mind. When applied to AI output, it produces an awkward question: can the required "effort" be attributed to a machine, and if so, to whom does any resulting right belong?
Three candidate rights-holders emerge from the existing legislative text. First, the developer of the AI system – the entity that designed the underlying model. Second, the entity that trained or fine-tuned the model on specific datasets. Third, the user who provided the prompt or instruction that triggered the specific output. None of these candidates fits cleanly within the statutory concept of "author." Each has a plausible argument; none has a conclusive one.
Patent protection for AI-related inventions raises a parallel set of difficulties. Israeli patent legislation requires an "inventor," and the Rashut HaPatentim (Israeli Patent Authority) has not formally accepted AI systems as inventors. Applications listing an AI as inventor have been rejected on doctrinal grounds consistent with the approach taken in the United Kingdom and the United States. The human-inventor requirement creates a practical problem for organisations whose most commercially significant innovations emerge from AI-assisted discovery rather than identifiable individual human contributions.
Trademark and trade secret protection are less conceptually troubled by AI generation. A trademark protects a sign used to distinguish goods or services in commerce – it does not require the sign to have been created by a human. Trade secret law protects confidential business information, and the process by which that information was generated is largely irrelevant to its protectability. These two instruments therefore offer more immediate practical value to businesses seeking to secure AI-related assets under current Israeli law.
Competing interpretations and the gap between statute and practice
Israeli courts have not yet issued a landmark ruling on AI-generated copyright. Practitioners in Israel note that the courts' existing doctrinal toolkit pulls in two directions. One line of reasoning, drawing on the purposive approach common in Israeli judicial method. Holds that copyright should vest in the human who made the most significant creative decision in the chain leading to the output. Under this view, the user who crafted the prompt – and exercised genuine creative judgement in doing so – has the strongest claim.
A competing interpretation, more cautious about expanding rights beyond statutory text, holds that where no human made a sufficiently original contribution to the specific output, the work simply falls into the public domain. This outcome is commercially unattractive to any business that has invested in building an AI system, but it follows logically from a strict reading of the human-authorship requirement.
A third position, less frequently aired but worth noting. Draws an analogy to the treatment of computer-generated works in UK legislation. This assigns copyright to the person who made the necessary arrangements for the work's creation. Israeli courts are not bound by English authority, but they do engage with common law jurisdictions in comparative analysis. The "arrangements" model has intuitive commercial appeal. It has not, however, been adopted in Israeli judicial decisions or endorsed by the legislature.
The gap between statute and practice is most visible in technology licensing. When an Israeli technology company licenses AI-generated software, images, or text to a foreign buyer, both parties typically assume that the licensor owns the licensed material and has the authority to grant rights. That assumption rests on no firm statutory foundation under current Israeli intellectual property legislation. A sophisticated buyer's counsel will increasingly identify this gap during due diligence. The absence of a clear chain of title can delay or derail transactions.
Software liability adds a further layer of complexity. Where an AI system generates code that causes loss. through a security flaw, an algorithmic error. Alternatively. An unintended interaction with existing systems. the question of who bears responsibility sits at the intersection of intellectual property doctrine and software liability principles. Israeli civil legislation provides general rules on liability for defective products and for professional negligence, but neither maps cleanly onto the behaviour of an autonomous generative system. Algorithmic accountability remains a theoretical concern in Israeli regulatory discourse rather than a settled area of law.
To explore how these issues compare in a neighbouring jurisdiction, our parallel analysis of AI-generated works and intellectual property in the UAE examines the DIFC and onshore frameworks, which face structurally similar doctrinal gaps.
Training data, AI Act compliance, and the regulatory horizon
A problem that receives less attention than authorship – but that carries equal commercial risk – is the legality of the training datasets used to build AI systems deployed in Israel. Israeli intellectual property legislation does not currently contain an explicit text-and-data-mining exception. The absence of such an exception means that training an AI model on copyright-protected works scraped from Israeli or foreign sources may constitute infringement unless a valid licence exists or a fair-use defence applies.
The fair-use doctrine in Israeli copyright law is a flexible, multi-factor test. Courts weigh the purpose and character of the use, the nature of the copyrighted work, the amount used, and the effect on the market for the original work. Applying these factors to AI training is genuinely unsettled. Practitioners point out that the transformative nature of training – where the model learns statistical patterns rather than reproducing specific content – may favour a fair-use finding. Equally, where training uses the entirety of commercially valuable datasets and produces outputs that substitute for those originals, the fair-use defence weakens considerably.
Beyond domestic Israeli law, international businesses must consider the extraterritorial reach of foreign AI regulation. The EU's AI Act compliance obligations apply to AI systems placed on the EU market, regardless of where the developer is based. An Israeli company exporting AI-powered digital services into Europe must therefore map its practices against the EU's tiered risk system and its transparency requirements. This creates a dual compliance burden: satisfying Israeli intellectual property and technology licensing requirements domestically, while simultaneously meeting AI Act compliance standards for European customers.
Israel does not yet have comprehensive AI-specific legislation. The Israeli Innovation Authority and relevant ministries have published policy papers and initiated consultation processes on AI governance. However. As of the time of writing no binding AI regulatory regime comparable to the EU's has been enacted. The regulatory trajectory points toward increased oversight. Businesses that build compliance infrastructure now – particularly around algorithmic accountability, documentation of training data provenance, and technology licensing terms – will be better positioned when binding rules arrive.
For a comprehensive view of the AI and technology law services available to businesses operating in Israel, see our practice overview of AI and technology law in Israel.
To explore an expert assessment of your AI-related IP exposure in Israel, contact us at info@ferrazwhitmore.com.
Cross-border implications for Asia-Pacific and Middle East clients
Clients headquartered in Singapore, Japan, the UAE, or Saudi Arabia who deploy AI systems in or through Israel face a set of compounding cross-border risks. The first is jurisdictional uncertainty over which country's intellectual property legislation governs AI-generated content. Where the AI system is trained in one country, operated in another, and its output used in a third, the connecting factors that determine the governing law are rarely obvious.
Private international law principles applied by Israeli courts generally look to the place of creation or first publication as a primary connecting factor for copyright disputes. For AI-generated works, that factor may be difficult to determine: generation happens instantaneously across distributed infrastructure, and "publication" in the traditional sense may not occur at all where content is used internally. Practitioners in Israel advise that explicit choice-of-law clauses in technology licensing agreements are therefore more important than ever. Without them, the applicable legal regime may be decided by a court applying conflict-of-laws rules that were not designed with AI in mind.
The second cross-border risk concerns the enforceability of IP rights in AI-generated works across multiple jurisdictions simultaneously. A business that secures the strongest available protection under Israeli law. whether through trade secret arrangements, trademark registration. Alternatively. Contractual restrictions in its technology licensing agreements. cannot assume that equivalent protection will be recognised in other countries. The UAE, for example, has begun to address AI-generated works through amendments to its copyright legislation, but the approach differs materially from the position in Israel. Japan has enacted specific provisions on AI-generated works that are more permissive than most other jurisdictions. Each market requires separate analysis.
The third risk is reputational and commercial rather than strictly legal. Cross-border transactions involving AI-generated content are increasingly subject to enhanced due diligence by institutional buyers, insurers, and regulators. A company that cannot demonstrate clean title to its AI-generated assets. or that has not documented its training data provenance and technology licensing chain. will find that counterparties in Singapore. Tokyo. Alternatively, Abu Dhabi apply additional scrutiny. The lost opportunity cost of delayed or failed transactions can substantially exceed the cost of building a compliant IP structure from the outset.
For businesses active across both Israel and the broader Middle East and Asia-Pacific region, the intellectual property strategy for AI-generated works should be designed with all relevant jurisdictions in view. That means assessing not only current Israeli intellectual property legislation but also the trajectory of regulatory change across the region. It also means structuring technology licensing arrangements to allocate ownership, risk. Additionally, indemnity clearly between developers. Trainers. Additionally, end users. so that no party is left without a remedy if the underlying IP position is later challenged.
Our practice in intellectual property in Israel advises clients on the full spectrum of protection strategies available under current law, with a particular focus on technology-intensive sectors where AI-generated content is commercially significant.
For a tailored strategy on AI-generated IP protection and technology licensing in Israel and across the Middle East and Asia-Pacific region, reach out to info@ferrazwhitmore.com.
Strategic recommendations and outlook
Businesses that wait for definitive legislative or judicial clarity before addressing their AI-generated IP position are making a costly choice. The legislative process in Israel – as in most jurisdictions – moves more slowly than AI development. A company that deploys AI systems today and defers IP structuring until the law is settled may find, several years from now. That its most commercially significant assets are legally unprotected or that its freedom to operate is constrained by claims it did not anticipate.
The following strategic considerations reflect current best practice in Israeli intellectual property and technology law, adapted to the specific challenges of AI-generated works.
- Document the human creative contribution at every stage of AI-assisted content production. The stronger the evidence of human judgement – in prompt design, curation, selection, and refinement – the stronger the authorship claim under current doctrine.
- Treat AI-generated works as trade secrets where possible. Trade secret protection does not depend on authorship and can be maintained indefinitely, provided confidentiality is actively managed.
- Audit training data provenance before commercialising AI-generated output. Identify which datasets were used, whether licences were obtained, and whether a fair-use defence is genuinely available.
- Draft technology licensing agreements that explicitly address AI-generated content: who owns it, who indemnifies whom for IP claims, and which jurisdiction's law governs disputes.
- Monitor Israeli legislative and regulatory developments. The period of consultation currently underway is likely to produce binding rules within the medium term. Businesses that have engaged with the consultation process – and that have already structured their operations to meet likely requirements – will adapt more smoothly.
The outlook for Israeli AI legislation points toward greater specificity rather than simple extension of existing copyright doctrine. Policymakers have signalled awareness of the economic significance of Israel's technology sector. Any legislative solution is therefore likely to balance the interests of AI developers and users against those of human creators whose works are used in training. The precise balance is not yet determined. What is clear is that businesses that have invested in AI-generated content as a commercial asset should treat the current period as an opportunity to build a defensible legal position. not as a reason to delay action.
Frequently asked questions
Q: Can an AI system hold copyright over a work it generates in Israel?
A: Under Israeli intellectual property legislation, copyright vests only in human authors. An AI system cannot hold rights. The question of who among the developer, trainer, or user acquires any rights remains unsettled by statute and has not yet been authoritatively resolved by the Israeli courts.
Q: How long does it take to register a copyright or software-related IP right in Israel?
A: Israel does not operate a formal copyright registration system. Rights arise automatically upon creation. Patent applications covering AI-related software inventions typically take several years to examine. Trademark registration at the Israeli Registrar of Trade Marks generally completes within one to two years, subject to opposition proceedings.
Q: Is it a misconception that training an AI on third-party content is always permissible in Israel?
A: Yes. A common misconception holds that training an AI model on publicly available content does not infringe copyright. Israeli intellectual property legislation does not currently contain an explicit text-and-data-mining exception comparable to those in some EU jurisdictions. Practitioners advise that training on copyright-protected material without a licence carries meaningful infringement risk until the legislature or courts provide clearer guidance. Engaging a lawyer in Israel with cross-border AI experience is advisable before commercialising AI-trained systems at scale.
About Ferraz & Whitmore
Ferraz & Whitmore is an international law firm based in Lisbon, advising business clients across 46 jurisdictions. As an international law firm operating across Israel and the broader Middle East and Asia-Pacific region. We combine 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. Our team advises technology companies, institutional investors, and in-house legal teams on AI Act compliance, algorithmic accountability, software liability, and the full spectrum of digital services regulation. The firm's AI and technology law practice covers jurisdictions across Europe, the Middle East, and Asia, supported by a network of local counsel in each market. Our attorneys have advised on AI-related IP matters across both civil law and common law systems, including matters before Israeli regulatory authorities and in cross-border technology transactions. To discuss your situation with a lawyer experienced in Israel and the wider region, 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.