In today’s connected world, the way we create, share and protect digital assets is no longer optional – it is a strategic imperative. From university staff drafting lecture slides to government offices processing citizen applications, three inter-linked themes dominate the conversation: digital accessibility the maturity of information within automated services, and the emerging convergence of intellectual property, data and AI regulation. Understanding each component, and how they reinforce one another, helps organisations reduce barriers, boost efficiency and safeguard innovation.
Embedding accessibility into everyday academic work
Higher-education institutions in the United Kingdom operate under the Equality Act 2010 and the Public Sector Bodies (Websites and Mobile Applications) (No. 2) Accessibility Regulations. These statutes obligate staff to produce content that meets the Web Content Accessibility Guidelines (WCAG) 2.2 Level AA the benchmark that defines four core principles – perceivable, operable, understandable and robust. While legal compliance is mandatory, true inclusion emerges from habits that staff can adopt without expensive tools. Simple choices—such as selecting high-contrast colour palettes, adding descriptive alt-text that explains an image’s purpose rather than its appearance, and using proper heading structures in Word or PowerPoint—make documents navigable for screen-reader users. Writing emails in plain English, crafting meaningful subject lines, and avoiding generic link text like “click here” further support comprehension. Sans-serif fonts and the avoidance of italics improve legibility, while built-in accessibility checkers provide a quick sanity check before publishing. By treating these actions as routine, much like confirming recipients before hitting “send,” institutions gradually embed an inclusive culture.
From scanned pages to connected knowledge: the Document-to-Knowledge framework
Public administrations increasingly automate evidence processing to cut costs, accelerate decisions and limit human error. Yet the effectiveness of such automation hinges on the refinement of the underlying information not merely on the software selected. The Document-to-Knowledge (D2K) framework, piloted in the GovTech4All “AI4Evidence” program, defines six maturity levels. A scanned image is merely a picture; an unstructured text extraction loses document hierarchy; structured text restores headings and lists; labelled information tags entities like dates or amounts; modeled information expresses relationships (e.g., person → address); and connected knowledge links these data points to trusted registers. Each ascent unlocks new capabilities—early levels enable basic sorting, while higher tiers support reliable rule-based eligibility checks and future “once-only” data exchanges across agencies. By answering three questions—where the data sits today, where it must be for current objectives, and where it could evolve—project planners clarify scope, avoid short-term fixes that damage long-term reusability, and set technology-agnostic requirements before any vendor is engaged.
Legal convergence: when copyright, database rights and trade secrets overlap with AI
Historically, intellectual-property regimes—trademarks, patents, copyright, database rights and trade secrets—operated in silos. Recent case law and legislation show that a single asset can trigger protection under several regimes simultaneously. In Portugal, Article 194 of the Industrial Property Code allows a registered design to enjoy concurrent copyright protection, while EU case law (Cofemel C-683/17) confirms that designs meeting the originality threshold may also be protected as works. This doctrinal blending extends to data: a curated dataset can be a copyrighted creative work a sui generis database protected for substantial investment, and a trade secret if kept confidential. The EU Digital Single Market Directive, transposed by Decree-Law 47/2023, introduced explicit text-and-data-mining exceptions, letting researchers mine lawfully available works unless rights-holders lodge machine-readable opt-outs. Consequently, publishers and creators must manage licensing, opt-out mechanisms and confidentiality clauses within a single contract, ensuring that AI-training pipelines respect layered IP rights while avoiding inadvertent infringement.
When organisations align inclusive design, high-quality information structures and a nuanced IP strategy, they create a virtuous cycle. Accessible content feeds cleaner data, which in turn simplifies compliance with evolving IP and AI regulations. Conversely, neglecting any of these pillars reintroduces barriers—whether for users with disabilities, automated decision-making systems, or innovators seeking to build on existing knowledge. The path forward is clear: embed accessibility into daily practice, assess information maturity early in any automation effort, and adopt an integrated IP-management approach that anticipates the overlapping rights of modern digital assets.
