Morning Program: Think Twice Conference
Afternoon Program: Digital Fairness Act
Digital Integrity, Decentralized Economy, Democratic Confederalism: Three Axes Against the Surveillance State of AI
Abstract
AI governance is framed as a regulatory question. It is also a constitutional one. The systems that train, host and deploy AI operate inside the architecture of surveillance capitalism. They feed on personal data treated as raw material, they centralize compute and decision-making, and they make public administrations dependent on foreign hyperscalers. Regulation alone cannot repair what this architecture erodes.
We propose another starting point. Personal data are not a commodity. They are constitutive elements of the person, like organs of the body, with inalienable rights attached. From this premise follows a new fundamental right anchored in the right to life: the right to digital integrity. Six operational components: protection against non-consensual processing, security in the digital space, the right to an offline life, the right to be forgotten, the right not to be measured, analyzed or surveilled, the right not to be judged by a machine. Geneva and Neuchâtel Cantons have enshrined it in their constitutions, with massive popular majorities. Europe can follow.
Digital integrity does not stand alone. It needs an economic and political infrastructure. We present three axes that work together. The right itself, with concrete legal consequences for AI: data minimization, anonymity by default, refusal of mass profiling, end-to-end encryption as a public good. Second, a decentralized network economy built on open protocols that enable triple-entry accounting. Third, democratic confederalism with strict subsidiarity, giving local communities the power to set their own rules.
Together, these three axes form a viable alternative to the digital dystopia. They are not utopia. They are being built by communities, entrepreneurs and constitutional drafters. AI governance can reinforce surveillance, or join this construction. We argue for the second.
Presentation title TBC
Abstract
Governments across the world are now deploying AI to automate services, support decisions, and manage public resources. Yet the frameworks guiding these choices are too often built not on evidence, but on myth. This talk, based on the book AI Tools, Not Gods (BTF Press, 2026), argues that before we can govern AI well, we must first confront the powerful stories that govern us.
AI discourse is currently shaped by four interlocking mythologies. The Magic myth anthropomorphises AI, cloaking it in wonder to deflect accountability. The Madness myth oscillates between gold-rush hype and paralyzing moral panic, producing governance whiplash. The Heaven myth frames AI as a salvation engine, encouraging democratic abdication in favour of tech prophets. The Sin myth casts the algorithm as a digital scapegoat, letting the human decision-makers who design, deploy, and profit from these systems off the hook.
Each myth generates what the book calls “policy hallucinations”: laws that govern shadows while real harms grow. Billions are wasted legislating for science-fiction scenarios. Concrete dangers such as biased public systems, opaque automated decisions, and concentrated power are systematically misdiagnosed.
The talk offers a practical antidote: a technology-neutral governance blueprint grounded in three principles: Accountability, Proportionality, and Durability. Treating AI as a product (with clear liability), governance as a public health challenge (with risk-based regulation), and regulation as principles-based law (capable of outlasting any particular technology) can replace myth-driven policy with democratic realism.
For a conference exploring AI’s impact on governance and digital freedom, this presentation makes the case that the most urgent reform is not technical. It is epistemic. We must stop governing by metaphor and start governing by evidence, before the myths govern us.
From Plasters to Policy: What Should AI in Public Services Actually Be For?
Abstract
Across Europe, welfare systems face an uncomfortable future with fewer
workers, growing needs and increasing demands for documentation,
transparency and accountability.
Consider a school nurse treating a child’s scraped knee. The bandage may
take a minute whilst documenting it can take considerably longer.
The documentation matters, accountability should not disappear. AI can
reduce the friction, for example by turning a short spoken note into a
medical-record draft for the nurse to review.
The important question, however, is not how many minutes AI saves. It is
what those minutes are used for afterwards.
If every efficiency gain simply creates room for more administration,
public services become faster without necessarily becoming better.
And if governments do not decide what AI is for, that purpose will
increasingly be shaped by procurement, vendors and whatever outcomes are
easiest to measure.
This talk argues for a clearer why. I would like to see a future
where AI returns time, attention and knowledge to the public mission while
preserving accountability and human judgement.
Presentation title TBC
Abstract
Generative AI increasingly shapes how we access information and create content. Despite that, there is little agreement on how its outputs should be governed. Laws are often vague, corporate practices opaque, and governments pull in different directions.
This session presents findings from a two-year global research project led by The Future of Free Speech at Vanderbilt University, conducted with a team of fifteen legal, policy, and technical experts. The project offers concrete evidence on how governments and companies govern AI-generated speech. The session explores whether vague provisions in the EU AI Act and Digital Services Act risk restricting lawful speech; how “culture wars” are influencing AI policy in the US; and how leading AI models actually handle speech in this geopolitical environment, based on empirical tests.
The talk proposes a framework grounded in international human rights law, which builds on the tests of legality, legitimacy, and necessity. Policymakers, companies, and researchers can use it to evaluate AI governance and promote laws and policies that protect freedom of expression while addressing genuine harms.
Human-Centric AI Governance: The Missing Layer in Responsible AI Adoption
Abstract
As AI governance becomes a boardroom priority, many organizations assume
that policies, principles, and oversight structures are enough to ensure
responsible adoption. Yet governance only works when it is understood and
enacted at the frontline, where employees make daily decisions, interpret
guidance, and interact with AI-enabled systems.
Too often, responsible AI remains a leadership conversation while
ground-level employees are left unclear about the purpose, expectations,
and practical implications of the policy.
This talk argues that the real challenge in AI governance is not only
whether organizations have rules, but whether people across the
organization can understand, trust, and apply them. Moving from fear to
fluency requires more than communication; it requires readiness.
Drawing on doctoral research, the presentation argues that organizations
should treat AI readiness as a governance issue rather than a
change-management afterthought. The proposed dimensions include
psychological safety, intercultural contextualisation, regulatory
compliance awareness, skill confidence, change resilience and leadership
AI readiness.
Fake Crowds, Real Money: AI, Monetisation, and the Simulation of Public Opinion
Abstract
Generative AI has made it cheap to simulate public opinion at scale. Political actors, commercial entities, and malicious networks can now produce massive volumes of text, images, and video that convincingly imitate human language, emotion, and engagement, making authentic participation increasingly hard to distinguish from coordinated manipulation.
Crucially, this is now a business model. Platform monetisation rewards visibility and virality, not accuracy or authenticity, so AI-generated content is deployed to maximise clicks, ad revenue, and algorithmic reach, regardless of societal cost. The result: public debates artificially amplified, suppressed, or distorted; perceptions of public opinion manufactured; and trust in democratic participation steadily eroded.
This session explores how we got here and why the response requires regulating monetisation systems and platform incentives, not just the content itself.
Death of Traits – Upstream AI Governance for a Diverse Future
Abstract
This conference asks who controls the data behind AI. This talk asks an earlier question: who decides what counts as data at all?
Today’s leading AI models learn mostly from English‑language content already online. Knowledge that was never digitised — such as customary law, community dispute resolution and local ecological knowledge — never reaches them. Our governance tools cannot see this gap: they test what AI produces, not what was missing from the start. This talk calls the result the Death of Traits.
Drawing on the EU AI Act, the talk sets out three practical answers: public documentation of training‑data choices, a community‑governed data fund, and cooperation between countries of the Global South. It also poses a hard question for open‑knowledge advocates: is open always fair?
Democracies in the Age of AI: AI’s Impact in Online Information
Abstract
The presentation seeks to establish a connection between the Global State
of Democracy and the role of AI in shaping the online information
environment.
It will examine the global state of democracy and its decline,
particularly in areas such as freedom of the press and freedom of
expression.
It will then explore the role of social media platforms, the attention
economy, and their embedded AI systems in disseminating online
disinformation worldwide, including AI-generated content, algorithmic
recommender systems and online advertising.
The presentation will also consider the geopolitical agenda of Big Tech
companies and documented cases where online platforms amplify
radicalisation. It concludes with recommendations for political parties
and policymakers navigating the challenges posed by AI, disinformation
and their impact on democracy.
Who Can Say No? Preserving Human Authority in Europe’s AI Future
Abstract
An AI system recommends denying someone a public benefit. A human
approves it. But did that person have the evidence, time and authority
to challenge the recommendation—and can the affected person get a
meaningful review?
This session examines the gap between placing a human in a workflow and
giving that person meaningful authority. Building on the EU AI Act’s
human oversight provisions for high-risk systems, it asks how oversight
can be tested under the pressures of everyday public administration.
A hypothetical public-service case will show how unclear responsibility,
overloaded reviewers and ineffective intervention can undermine
accountability even when a human formally signs off.
Drawing on the WHEEL research programme, the presentation introduces
practical questions for policymakers, public-service leaders and civil
society: Who may challenge the decision? What evidence can they access?
Can they pause or reverse the action? What remedy is available to the
person affected?
The aim is to support AI adoption that strengthens public services while
preserving democratic accountability and digital freedom.