In today’s world, data is copied, shared, and reused at an unprecedented speed. A single dataset can travel across continents in seconds, be integrated into new products, or quietly shape decisions made by algorithms that affect millions of people. Yet despite this immense value, there is still no reliable way to answer a simple question: Who owns this data, and who is responsible when it is misused?

For centuries, societies have developed tools to protect ownership of creative work. We have copyright laws, watermarks on paintings, signatures on documents, just to name a few. But as more of our economy moves into the digital realm, these traditional safeguards struggle to keep up. Digital files and images can be duplicated perfectly and endlessly. A dataset shared with one partner today might resurface years later in the hands of someone entirely different. And when data is leaked, stolen, or misappropriated, tracing responsibility is a big challenge. Companies, researchers, and public institutions hesitate to share valuable data, fearing a loss of control. Citizens worry about how their information is handled. Regulators face growing pressure to enforce digital rights but lack practical mechanisms to verify who had which data and when.

Digital watermarking to the rescue

Instead of trying to lock data away, data watermarking embeds an invisible “signature” directly into the dataset itself. Each shared copy carries a unique marker that identifies its rightful owner and the specific recipient it was shared with. If the data shows up somewhere unexpected, the watermarks offer a capability that can reveal the owner, the source, and the path the data took. In other words, it enables traceability, a core requirement for a trustworthy data ecosystem.

The idea echoes historical practices: cartographers once marked maps to detect copyright violations [1], and paper manufacturers marked their produced paper as an early form of trademark [2]. But applying similar ideas to digital data requires new technical imagination. Digital watermarking must survive transformations, usage, and analysis, and it must integrate with legal frameworks addressing the intellectual property rights [3].

Figure 1. Watermarks through the history: antique paper presumably made in the 16th century in Italy; trap streets on an LA map by Thomas Brotehter Map Co. in 1966; digital watermark on a photograph by shutterstock.com

Why does IP protection matter?

Data sharing powers everything from medical research to media, from AI development to public administration. Without mechanisms to ensure responsible use, society risks either over-sharing (and losing control) or under-sharing (and losing innovation). Data watermarking contributes to a future where data can be shared confidently, where misuse can be detected, and where digital resources remain trustworthy.

Solid, transparent, and technically sound solutions are crucial. Data watermarking has already become one of the key building blocks of this next phase: a safer, fairer, and more accountable digital world.

This dissertation, therefore, strives to improve the robustness and usability of Intellectual Property Protection technologies and aims to provide their full open-source code for facilitating future development. 

[1] Alexander, I., 2023. Copyright and Cartography: History, Law, and the Circulation of Geographical Knowledge (p. 352). Bloomsbury Academic.

[2] Müller, L., 2021. Understanding paper: Structures, watermarks, and a conservator’s passion. Harvard Art Museum.

[3] Graux, H., 2024. What is Data Ownership, and Does it Still Matter Under EU Data Law?: An Exploration of Traditional Concepts of Data Ownership, and of the Expected Impact of the Data Act. Publications Office of the European Union.