UKRAINE Law and Practice Contributed by: Yaroslav Ognevyuk, AMBASSADORS
7. Data, AI and Emerging IP Issues 7.1 Data Rights and Database Protection Ukrainian law does not recognise a sui generis data- base right comparable to the EU database regime. Protection of datasets is based on a combination of copyright, contractual arrangements, and trade secret law. Copyright may apply to the structure or selection of a database where it reflects sufficient originality. The underlying data is not protected as such. Businesses, therefore, rely on contractual frameworks to regulate access to, use of, and redistribution of datasets, par- ticularly in technology- and platform-driven environ- ments. Trade secret protection is available where datasets are not publicly accessible, have commercial value, and are subject to confidentiality measures. This approach is common in data-driven sectors, including AI and analytics. Scraping and data extraction are not regulated as standalone concepts. Lawfulness is assessed through existing legal frameworks, including contractual terms of use, unfair competition rules, and, where applica- ble, data protection considerations. 7.2 AI-Generated Works and Inventorship/ Authorship Ukrainian law is built on the premise that authorship and inventorship belong to natural persons. This prin- ciple directly shapes the treatment of AI-generated outputs. Copyright protection is generally available only where a human author can be identified as having made a creative contribution. Fully autonomous AI-generated outputs, where no meaningful human input can be demonstrated, fall outside the traditional scope of protection. Qualification depends on the degree of human involvement in prompting, selecting, and refin- ing the output. A similar approach applies in patent law. AI cannot be recognised as an inventor. Applications must desig- nate a natural person, and current practice does not
permit AI systems to be named as inventors, regard- less of their role in the inventive process. This creates a structural mismatch between legal attri- bution and technological reality. Rights are typically allocated to employees or contractors who design, train, or operate AI systems, with contractual provi- sions governing outputs generated with AI assistance. Ownership is secured not at the level of the algorithm itself, but through control over the process and the resulting outputs. Clear internal policies and contrac- tual frameworks are critical, particularly where multiple actors contribute to AI-assisted creation. 7.3 Training Data, Model Development and Infringement Risk The use of third-party content in training datasets raises a combination of copyright and trade secret risks under Ukrainian law. The primary issue is not the training process itself, but the origin and status of the data being used. These risks are increasingly relevant for companies developing AI models in data-intensive environments. Where copyrighted works are included in training datasets without authorisation, this may constitute infringement, particularly where the data is repro- duced, stored, or further distributed. Ukrainian law does not provide a clear or broad exception for text and data mining comparable to the EU framework. Reliance on implied permissibility is therefore risky, especially in commercial settings. Ukrainian law lacks a dedicated statutory framework for text and data mining, increasing legal uncertainty for commercial AI development. Trade secret exposure presents a separate risk. If train- ing data includes confidential information obtained through employees, contractors, or third-party inte- grations, liability may arise even without direct copy- ing. The key question is whether the information was accessed and used in breach of confidentiality obli- gations. Enforcement remains in development but follows exist- ing IP mechanisms. Rights holders may seek injunc-
384 CHAMBERS.COM
Powered by FlippingBook