Intellectual Property in the Age of AI: Protecting Data Value

Artificial intelligence has reshaped how organisations create, process, analyze and commercialise information. From generative models trained on vast collections of digital content to AI-powered research and product development, data has become one of the most strategic assets in the global economy. Yet, as businesses accelerate AI adoption, a fundamental question remains: how can organisations unlock the value of data while protecting the intellectual property, privacy and commercial interests embedded within it?

Governments, regulators, technology providers and businesses are reconsidering how intellectual property rights should operate in a world where machines can learn from existing works and generate new content. For enterprises, this is no longer simply a legal discussion. It is a practical challenge involving data access, security, governance and the ability to turn information into measurable business value.

The organisations best positioned to benefit from AI will not necessarily be those with the largest datasets or the most powerful models, but those capable of using data responsibly, securely and controlling how information is processed, shared and reused. This is where data infrastructure becomes a strategic component of intellectual property protection and responsible AI adoption.

A Global Intellectual Property Framework in Transition

The relationship between intellectual property and artificial intelligence is evolving across jurisdictions, with different legal systems exploring how existing protections should apply to emerging technologies.

At the international level, the World Intellectual Property Organization (WIPO) continues to provide a forum for discussions about copyright, patents, inventorship, ownership and the use of protected material in AI development. Its work reflects the growing need to connect legal principles with the technical systems supporting rights management and intellectual property processes.

The European Union is also developing its approach. The EU AI Act establishes obligations for providers of general-purpose AI models, including requirements concerning technical documentation, copyright policies and public summaries of training content. These provisions strengthen the connection between AI development, transparency and copyright-related responsibilities.

In the United Kingdom, the government published a report and impact assessment on copyright and AI in March 2026, examining possible approaches to balancing the interests of rights holders, AI developers and users. Meanwhile, in the United States, the Copyright Office has examined both the copyrightability of AI-generated outputs and the use of protected material to train generative models. Its work highlights the importance of human authorship and the complexity of applying existing copyright principles to AI.

Taken together, these developments reveal a global landscape in transition rather than a single, settled international standard. Businesses must navigate different legal requirements and continuing uncertainty while making operational decisions today.

For organisations operating across borders, the challenge is twofold: understanding the applicable legal framework and building technical processes capable of supporting compliance as requirements evolve.

The Intellectual Property Paradox of Artificial Intelligence

AI creates a paradox for organisations that want to innovate. The more valuable their data becomes, the more difficult it can be to make that data available for experimentation, collaboration and commercial use.

A financial institution may possess transaction histories that could improve fraud detection. A healthcare organisation may hold information that could support research, while an industrial company may have years of operational data that could improve forecasting and production efficiency. In each case, valuable information may also contain personal data, confidential knowledge, copyrighted material, trade secrets or information subject to contractual restrictions.

The result is a tension between innovation and protection. Keeping data inaccessible limits its potential value, but making it broadly available can expose organisations to privacy violations, unauthorised disclosure, contractual disputes or the loss of competitive advantage.

The problem becomes more complex when information moves between organisations or is processed by external AI systems. Who can access it, for what purpose, and under which conditions? Can it be used to train a model? Can the resulting outputs or datasets be reused commercially? Can the organisation demonstrate that appropriate permissions and controls were in place?

These questions require more than a privacy policy or a model. They demand a connected approach to data preparation, protection, access control, traceability and governance.

It is also important to distinguish between different forms of protection. Copyright, trade secrets, contractual rights, data protection and intellectual property ownership are related, but not interchangeable. Anonymisation does not automatically grant permission to use copyrighted material, and synthetic data does not eliminate every legal or confidentiality risk. Effective protection requires the appropriate combination of legal assessment, technical safeguards and operational controls.

Why Traditional Data Protection Is No Longer Enough

Many organisations approach data protection as a preliminary compliance step: information is collected, access is restricted and a review takes place before a project begins. However, data is continuously ingested, transformed, enriched, analysed and shared, and its risk profile can change at every stage.

A dataset that appears harmless in isolation may reveal sensitive information when combined with another source. Information suitable for internal analysis may not be appropriate for external sharing, while models trained on proprietary data can raise questions about confidentiality, ownership and permitted use.

Protection therefore cannot be treated as a one-time intervention. Organisations need to understand their data, identify sensitive content, apply suitable privacy measures and define how information can be accessed and reused. They must also maintain sufficient traceability to demonstrate that appropriate controls have been applied.

This is particularly important in regulated environments, where data governance must support accountability, auditability and risk management.

The strategic objective is not to prevent data from being used. It is to make responsible data use possible without surrendering control.

From Restricted Information to Usable Data

One promising approach is to separate the value of information from unnecessary exposure of the original records.

Anonymisation and pseudonymisation can reduce exposure when appropriately implemented, although they provide different levels of protection. Pseudonymised data may remain personal data under applicable law, while anonymisation must be assessed against the realistic possibility of re-identification.

Synthetic data offers another approach. By generating artificial datasets that reproduce selected statistical properties or patterns of source data, organisations can support analysis, software development, testing and certain AI workflows while reducing reliance on original records.

However, synthetic data is not automatically risk-free. Its value depends on whether it remains useful for the intended purpose, while its privacy characteristics depend on how it is generated and evaluated.

Data quality, privacy protection and governance must therefore work together. A dataset that is safe but analytically unsuitable will not deliver business value; one that is useful but inadequately protected may introduce unacceptable risks. Traceability is equally important when transformed data needs to be approved, audited or reused.

The goal is to create a controlled path from sensitive or restricted information to data that can be used for a clearly defined purpose.

How Dedomena.AI Supports the Data Protection Lifecycle

Dedomena.AI provides an integrated data and AI infrastructure designed to help organisations prepare, protect, govern and use data across connected workflows. Its approach brings together capabilities that can support different stages of the data lifecycle.

The first stage is understanding and preparing data. Data validation and quality processes help identify inconsistencies, duplicates, structural issues and other problems that can undermine analysis. This preparation supports not only AI training but also conventional analytics, reporting, research and secure data exchange.

The next stage is protecting sensitive information. Dedomena.AI's anonymisation and synthetic data capabilities support workflows in which organisations need to reduce exposure to original records or make alternative datasets available for analysis and development. The appropriate method depends on the data, its intended use and the relevant risk assessment.

Protection alone, however, is not enough. Data must also move between systems and organisations under appropriate conditions. Dedomena.AI's data space capabilities support secure, governed data collaboration, helping organisations make information available to authorised participants without treating unrestricted access to original data as the default.

Governance and traceability connect these activities. Policies, monitoring and records of data processes can help organisations understand how information is prepared and used, support oversight and improve the evidence available for review. Depending on the configured workflow, these controls can help embed governance into data operations rather than leaving it entirely to manual documentation.

For AI-related use cases, these capabilities can support dataset preparation, privacy measures and controlled data use. Where models or AI applications are involved, organisations can extend the workflow to relevant model and governance capabilities according to the use case. Dedomena.AI works as a secure and trusted data layer before the information hits the LLM or AI assistant.

Crucially, Dedomena.AI is not limited to scenarios involving AI training. The same foundations can support data quality, anonymisation and controlled access for non-AI purposes. For example, an organisation may need to anonymise CVs to reduce the exposure of personal information during a compliance process, without using the data to train a model.

This distinction matters because the broader challenge is not simply making data available to AI. It is making data usable for legitimate purposes while maintaining appropriate control over sensitive information.

Connecting Data Utility, Privacy and Intellectual Property

The value of an integrated approach becomes clearer when considered from a business perspective.

Imagine a financial institution collaborating with an external partner on a fraud-analysis initiative. It needs to understand the quality of its transaction data, reduce unnecessary exposure of sensitive records and establish conditions for controlled access. If model development is involved, the parties may also need to evaluate data suitability and document relevant governance processes.

When these activities are managed through disconnected tools and manual handovers, organisations can face duplicated work, inconsistent controls and limited visibility. A connected workflow can help bring data preparation, protection and collaboration into a more coherent operational framework.

The same principle applies to proprietary research, internal documents, industrial information and other commercially sensitive assets. Before making information available for analysis or collaboration, organisations need to understand what it contains, determine the appropriate protections and establish how it can be used.

Dedomena.AI supports this technical foundation by bringing data preparation, privacy-focused processing, synthetic data and governed data sharing into a connected environment. This can help organisations operationalise their data policies and establish more consistent processes for responsible data use.

Technology nevertheless has clear limits. A platform cannot independently determine the legal ownership of every dataset, settle copyright disputes or grant permissions an organisation does not possess. Such questions may require legal review, licensing agreements and decisions by the relevant rights holders. Technical infrastructure can help implement those requirements, control access and maintain records that support oversight.

Data governance is therefore not a substitute for intellectual property management. It is one of the mechanisms through which organisations can put their rights, obligations and internal policies into practice.

Building a More Responsible Data Economy

As intellectual property frameworks evolve, organisations should avoid two extremes: treating every uncertainty as a reason to stop innovation, or assuming that technical access to data means it can be used without restrictions.

A sustainable approach begins with a clear understanding of the intended purpose, the rights and obligations associated with the information, the risks created by its use and the controls required throughout its lifecycle. Data quality, anonymisation, synthetic data, controlled collaboration and governance each address different parts of this challenge.

For business leaders, this is increasingly a question of competitiveness. Organisations that can establish trustworthy processes for using and sharing data may be better positioned to develop new services, improve decisions and collaborate with external partners. Those unable to demonstrate appropriate control may find that valuable information remains inaccessible or that the risks of using it outweigh the potential benefits.

The next stage of AI adoption will depend not only on advances in models and computing, but also on the infrastructure that makes data usable, protected and governable.

At Dedomena.AI, our focus is on enabling that foundation. By supporting data preparation and validation, anonymisation and synthetic data, secure data collaboration and governance across connected workflows, we help organisations move towards a more controlled and practical use of their information.

The future of intellectual property in the AI era will continue to be shaped by legislation, court decisions, licensing models and international cooperation. Businesses cannot wait for every question to be resolved before improving how they manage their data. They can, however, start building the technical and organisational foundations needed to use information responsibly today.

The real opportunity is not to choose between protecting data and creating value from it. It is to build the infrastructure that makes both possible.