Digital Integrity: A Challenge For Humanity In The AI Era

Type a product into a search bar, and in no time a plethora of similar products, along with related supplements and complementary items, will populate your promotional inbox and search engine suggestions. Similarly, search for a particular program or news topic, and the feed will fill with similar content. This is how the algorithm works: it collects data from digital footprints, studies patterns, and makes predictions to enable speed and convenience. This is probably the most succinct explanation of how an AI algorithm works.

In technical terms, that is exactly how machine learning works. But what does that mean for the human element? AI and algorithmic decision-making have become the central engine in so many areas, including healthcare, banking, higher education, government departments, and corporations.

The underlying issue is that when individuals’ digital footprints are aggregated, they feed demographic-behavior algorithms that can become powerful persuasive weapons. Rather than recommending content based solely on your personal browsing history, these models curate it using the aggregated habits of others who share your demographic profile.

This creates a severe integrity deficit. Users lose the ability to shape their own digital environment, as their feed is dictated by the average interests, biases, or vulnerabilities of a group of strangers.

This leads to AI manipulation, where automated systems rather than humans shape what a person sees and chooses based on predetermined patterns. This system-driven decision-making directly affects people’s lives. That, of course, leads to the fundamental question: does digital integrity even exist in an age of algorithms?

What is Digital Integrity?

Digital integrity is the framework that ensures our digital ecosystem remains honest, fair, and reliable. It requires accurate data feeding AI systems, transparent algorithms processing it, and outputs that do not erode human agency or democratic values. But achieving this integrity is one of the greatest challenges of today’s tech era.

Alexis Roussel, in his book “Notre si précieuse intégrité numérique: Plaidoyer pour une révolution humaniste” (OUR Precious Digital Integrity: A Plea for a Humanist Revolution), argues that data isn’t just a commodity for tech companies; it’s part of people’s identity. This highlights why protecting citizens in the digital world matters. From this view, digital integrity isn’t just a legal or technical issue; it’s about protecting human rights.

Concerns about the digital integrity of AI

The term “artificial” intelligence refers to how it is made. It works as an input-and-output model: data goes in, patterns are built, and the data shapes the results. So, the results depend completely on the quality of the data, which means data integrity is essential for digital integrity. In short, AI digital integrity isn’t just one thing, but a layered structure:

• Data Integrity (inputs)
• Model & Algorithmic Integrity (the processing)
• Information & Ecosystem Integrity (the outputs)

From a data-feeding/input perspective, AI is in a structural crisis (which researchers also refer to as representation bias and algorithmic homogenization).

Systematic Exclusion

The drive to improve AI output is actually making it less inclusive. Major AI models learn by reading the web, and the web doesn’t represent everyone. The web is overwhelmingly English. If a culture passes its history down through oral tradition rather than websites, or if a community speaks a distinctive local dialect, the AI simply does not see it. It ends up treating the Western, English-speaking way of being human as the default.

To clean up the internet’s messy data, tech companies hire human reviewers to filter out bad content. However, these reviewers are usually based in a few major tech centers like California or London. If a reviewer flags an unfamiliar African or Asian phrasing as “wrong” or “aggressive” simply because they do not understand the cultural nuance, that unique viewpoint is permanently erased.

This creates a dangerous loop: early AI models accidentally lock out smaller cultures. And because the internet is now flooded with AI-written text, newer models end up learning from the flawed work of older models. It is like a beloved family recipe being rewritten by a giant corporate factory. Each time the system updates, it tweaks the ingredients to suit a mass-market, standardized taste. In the process, the factory completely leaves out the rare local spices that made the heritage dish special. Eventually, the world forgets the original recipe ever had that distinct flavor. The result is a model that creates an “inbreeding loop” – treating minority perspectives as anomalies or errors and fundamentally undermining the soundness of the data.

Helen Lalthanpari, founder of the Data & AI Governance Institute (DAIGI), has thoroughly researched and clearly explained such architectural flaws in her paper, The Death of Traits: Epistemic Exclusion in AI.

When Institutional Decision-Making Is Outsourced – The Loss of Sovereignty

As explained earlier, when such fine-tuned data enters public infrastructure, it reinforces an unexplainable bias. As algorithms replace human monitoring and prioritize efficiency over equity, the threat shifts from cultural erasure to the erosion of democratic agency.

Jonatan Tensetti, an AI transformation lead and prominent voice for digital equity, discusses this exact risk in his book Med AI för en likvärdig skola (With AI for an Equitable School). When public bodies outsource administrative work to AI infrastructure, they open public data to closed systems. And when a decision is made, for example, an application for a health service is rejected based on preset, biased criteria, it leaves the public with no choice but to accept it without understanding and without the ability to challenge it.

If a citizen cannot understand or challenge how an automated public system has denied them a service, a benefit, or an opportunity, then that body’s digital integrity is effectively at an end.

Workforce Fluency and the Illusion of Human Control

Step back from the input and training side of things, and the human element on the periphery is under enormous pressure. Legislation is clear: there must be a “human in the loop” to preserve the model’s integrity, but its effectiveness remains in doubt. Genuine digital integrity rests with the people working the system day to day. To spot a biased or broken AI decision, a human needs proper training, time, and the psychological safety to speak up.

This is similar to a plane with autopilot. If a pilot relies too much on the computer, their flying skills get worse over time, a problem called capacity degradation. If the airline prioritizes quick departures and punishes pilots for delays, pilots will feel pressured to let the computer fly, even if they notice a small mistake.

Narmadha Kamalakannan, AI-readiness and human-centric transformation advisor and founder of HumIA5, explained this bottleneck in her study, From Fear to Fluency. Gauging AI readiness is the key here. When workforce members fear emerging systems or worry about career repercussions for challenging automated outputs, critical thinking diminishes. Employees essentially function like fatigued monitors, passively approving algorithmic results to meet organizational demands. Ultimately, systemic integrity fails when supervisors lack the psychological safety, preparation, or time to intervene and stop flawed decisions.

Need of the Hour – A Collective Redress: Think Twice 2026

Understanding the various layers of the automated system shows that the fundamental vulnerability of AI is not in the software per se but in the social dynamics. It’s in the failure to include the full range of cultural nuance, languages, and local heritages in the input streams; it’s in the surrender of control to a small number of private companies, which in turn threatens public rights; it’s in the institutional pressure that will see workers simply go along with the automated output being generated without any real oversight. The fundamental threat is the steady erosion of human agency.

From that perspective, the issues outlined above have to be addressed and resolved. A passive regulatory regime will not be adequate. You need active civic engagement around who is going to govern algorithmic power.

Think Twice 2026 (17 October in Brussels) focused on this core idea. Hosted by the European Pirates in conjunction with Pirate Parties International, it was an opportunity to get together and talk about the current state of the technology debate. Rather than looking at individual tools, the event sought to address the structural problems of the decade by looking at two interlocking issues: AI for governance and the governance of AI.

By bringing together people from different fields, such as researchers, technology experts, administrators, journalists, and public advocates, the event sparked a vital conversation across Europe. It shed light on critical concerns like tracking cultural bias, institutional control, workforce readiness, and more.

The Brussels dialogues reached a clear conclusion: society must stop merely reacting to new technology. The key issue now is to actively shape the development of new automated technologies, so that as computers get faster and more powerful, human choice and human independence remain at the heart of it.

Acknowledgments and Gratitude

On behalf of the European Pirates and Pirate Parties International, we extend our heartfelt gratitude to the remarkable speakers, panelists, moderators, and attendees who made Think Twice 2026 a resounding success.

We extend our sincere thanks to:

• Alexis Roussel (Co-founder of Nym.com & Co-author of “Our Precious Digital Integrity”)
• Belén Luna Sanz (Policy & Advocacy Strategist at WHAT TO FIX)
• Caroline De Cock (Head of Research at Information Labs & Author of “AI Tools, Not Gods”)
• Helen Lalthanpari (Founder of the Data & AI Governance Institute – DAIGI)
• John A. Kwarsick (Founder of Kwarsick Consulting & Creator of WHEEL)
• Jonatan Tensetti (AI Development Strategist at the Municipality of Trelleborg & Spokesperson for the Swedish Pirate Party)
• Jordi Calvet-Bademunt (Senior Research Fellow at The Future of Free Speech, Vanderbilt University)
• Narmadha Kamalakannan (Founder of HumIA5 & Doctoral Researcher at École des Ponts Business School)
• Sebastian Becker Castellaro (Associate Programme Officer for Digitalization & Democracy at International IDEA)

Join Us

Together, we can build a digital future anchored in integrity, transparency, and freedom. Join European Pirates in this movement to uphold freedom for everyone.