Protecting democracy from disruptions caused by artificial intelligence
- Author(s):
- Parliamentary Assembly
- Origin
- Assembly
debate on 24 June 2026 (23rd and 24th sittings) (see Doc. 16417, report
of the Committee on Political Affairs and Democracy, rapporteur:
Ms Deborah Bergamini). Text adopted by the Assembly on
24 June 2026 (24th sitting).Provisional version subject
to editorial review.
1. The development
of artificial intelligence (AI) technology is advancing at an unprecedented
rate, holding the promise of improving many aspects of human life
and accelerating social and economic development. Indeed, AI may
represent the most transformative revolution in human history. However,
efforts to ensure that AI systems are safe and regulated by democratic
governance frameworks are falling short of keeping up with the pace
of innovation.
2. The Parliamentary Assembly is deeply concerned about the potentially
disruptive impact of AI on democracy in Europe and beyond. At the
same time, the Assembly acknowledges that this technology should not
be demonised, but can be instrumental in innovating democratic systems,
provided that all stakeholders understand what is at stake and take
immediate action.
3. With the right governance framework in place, AI can be harnessed
to strengthen democratic processes and institutions, as emphasised
in the Roadmap Towards a New Democratic Pact for Europe. AI can
facilitate public participation by providing citizens with access
to information, explaining complex policies, mediating deliberations,
and identifying patterns and common positions among large groups
of people. On the other hand, it could also enable politicians and
public authorities to gather proposals and information about citizens’ aspirations.
In this sense, AI can be a powerful tool for participatory and deliberative
democracy.
4. AI can promote inclusiveness by eliminating socio-economic
barriers and providing disadvantaged groups with better access to
public services, education and job opportunities. Furthermore, AI
can enhance the protection of human rights, improve the efficiency
with which public administration delivers services to citizens and
detect the malicious use of other AI tools.
5. In this context, the Assembly believes that Europe should
not restrict itself to a regulatory role, while cutting-edge AI
developments are mostly taking place in the United States and in
China. Guided by the values of democracy, human rights and the rule
of law, Europe should take the lead in shaping the development of new
AI applications, placing strong emphasis on their human dimension.
The risks associated with AI technologies must be effectively identified,
addressed and mitigated. In this regard, the Council of Europe has a
critical responsibility to ensure that the relationship between
humans and AI remains firmly oriented towards the common good.
6. AI technology relies on large datasets to train its systems
and produce outputs. Sensitive personal data, often collected infringing
on citizens’ privacy, can be exploited by malevolent individuals,
companies or governments, for mass surveillance purposes, predictive
policing, risk and social scoring, and censoring political opinions.
This has chilling effects on civil participation.
7. The datasets used to train AI systems can be polluted by politically
based disinformation content. They can also contain biases, as they
might reflect or even amplify inequalities already present in societies
in the outputs produced. When AI is used for policy-making purposes,
these biases can lead to ill-informed decisions or discrimination
against certain groups, such as women or minorities. Additionally,
AI systems can sometimes “hallucinate”, generating incomplete or
misleading information.
8. Personal data can be misused to create fake identities or
produce synthetic content, such as deepfakes in the form of texts,
images, audio files or videos, which essentially impersonate others
for the purpose of harassing, scamming, blackmailing or committing
other frauds.
9. Deepfakes and other synthetic content can also be used to
spread fake news, hate speech, and divisive content. Even with limited
time and resources, well-co-ordinated disinformation campaigns can
be launched on a large scale to interfere with political debates
and elections. Through “bot farms” and “troll farms”, namely networks
that create fictitious profiles programmed to automatically spread
disinformation, these operations can reach a wide audience extremely
quickly, in stark contrast to the time and resources needed to “debunk” them.
10. Malevolent agents, including foreign actors, are increasingly
exploiting these opportunities as hybrid methods to interfere with
democratic processes and institutions in Europe. The Assembly therefore
welcomes the ongoing work of the Committee of Experts on Foreign
Information Manipulation and Interference, particularly with regard
to the feasibility study on the possible elaboration of a legal
instrument on foreign information manipulation and interference,
including disinformation.
11. An increasing number of citizens are relying on social media
platforms as their main source of information. In turn, data gathered
from citizens is also used to profile them, identify their preferences
and micro target them with specific content, ultimately with the
aim of manipulating their opinions. Micro targeting contributes
to the creation of echo chambers, in which users are exposed to
a limited range of ideas and beliefs and do not have the opportunity
to engage in a meaningful political dialogue with people who hold
different opinions. Chatbots, which are software designed to simulate
a conversation, can also be programmed to censor specific content
and provide biased answers, thereby further contributing to the
manipulation of citizens’ choices and beliefs.
12. The business model of large social media platforms aims to
monetise user engagement through advertisements. This leads to algorithms
prioritising controversial and polarising content. Although most
of these platforms include fact-checking and moderation tools, these
are not always efficient or quick enough to prevent dangerous content
from being disseminated. In addition, polluted information ecosystems
and biased AI outputs can erode citizens’ trust in traditional media
and in democratic processes and institutions, further contributing
to democratic backsliding.
13. In the near future, AI systems may be entrusted with more
decision-making responsibilities, which could lead to misjudgements
regarding AI capabilities, lower human cognitive standards, and
limit diversity of perspectives. New developments could even lead
to the creation of a “super” AI, which would surpass human abilities,
potentially develop its own consciousness and escape human control,
with disastrous consequences.
14. Controlling large datasets, energy grids, computing capacity
and the human skills needed to develop and deploy AI systems has
become a crucial strategic geopolitical asset. This control can
lead to dangerous concentrations of power in the hands of a few
private actors and to increased global inequalities and tensions between
competing States.
15. In line with these considerations, the Assembly affirms that
democratic oversight must be maintained throughout the entire lifecycle
of AI systems, from development to deployment, to safeguard human
dignity, transparency and accountability. This must be achieved
in collaboration with all the relevant stakeholders, starting from
the private actors operating in the field of AI, to ensure that
AI systems are truly human-centred and empower humans without replacing
them.
16. To this end, the Assembly urges member and observer States
of the Council of Europe to sign and ratify the Framework Convention
on Artificial Intelligence and Human Rights, Democracy and the Rule
of Law (CETS No. 225) in order to allow its rapid entry into force
and encourages non-member States worldwide to request to accede
to this international treaty, the first of its kind in this field,
as soon as they have the opportunity to do so.
17. Furthermore, the Assembly calls on member and observer States
of the Council of Europe to:
17.1 engage
in a comprehensive reflection on the need to regulate the development
and deployment of all new AI technologies, including those specifically
intended for national security and defence purposes, while at the
same time eliminating the barriers to innovation in Europe;
17.2 take active action in response to threats of foreign interference
and to strengthen resilience against disinformation and misinformation,
in line with Assembly
Resolution
2593 (2025) and
Recommendation
2292 (2025); this should include adopting specific measures to counter
the misuse of AI tools for spreading deepfakes, for example by introducing
mandatory content fact-checking, debunking and watermarking on social
media platforms;
17.3 strengthen the transparency, explainability, accessibility
and inclusiveness criteria required by AI technologies, including
those used by the public sector or to deliver public services, as
well as the algorithms and content moderation tools used by social
media platforms;
17.4 encourage the use of diverse languages in the development
of AI technologies, to ensure their diversity and representativeness;
17.5 support and promote the development of open-source environments;
17.6 ensure that AI developers, providers and social media
platforms are held accountable for any harm caused by their services;
17.7 establish clear, effective and appropriate redress measures
that are easily accessible to victims of harm caused by AI;
17.8 incorporate risk and impact assessment and mitigation
procedures throughout the entire lifecycle of the AI systems, focusing
on human rights, democracy and the rule of law, for example by involving national
human rights institutions, and adopting the Methodology for assessing
the risk and impacts of artificial intelligence systems from the
perspective of human rights, democracy and the rule of law (HUDERIA)
of the Council of Europe;
17.9 strengthen AI and media literacy at all levels of education,
including informal education, for all age groups (with a specific
focus on older people), to enhance pre-bunking skills, coding capabilities,
ethics, critical thinking and general knowledge of how AI technologies
work, ensuring that no one is left behind in the digital transition;
17.10 dedicate sufficient resources to ensure that civil society
and citizens participate in all debates around AI technologies,
and that all relevant stakeholders, particularly young people, participate
in co-developing new regulations, standards and oversight mechanisms;
17.11 tackle discrimination and bias against women generated
by AI and ensure equal opportunities for women and men in the development,
deployment and use of AI technologies;
17.12 dedicate resources to foresight and research into the
multi-layered impact of new AI technologies on individuals’ psychology,
information ecosystems, democratic institutions and processes, energy production,
the environment and society at large, to be able to respond whenever
necessary through regulation, oversight and other preventive or
mitigating measures;
17.13 define clear frameworks for their relations with big tech
corporations, so that their influence on democratic societies can
be properly scrutinised and regulated whenever necessary, including
through appropriate antitrust laws to prevent oligopoly formation
and concentration of power;
17.14 consider defining clear red lines on the development of
certain technologies and a moratorium on the development of super
AI;
17.15 ensure that there is no protection gap between approaches
to regulating AI and approaches to regulating online platforms with
respect to the increasing integration of generative AI tools into
social media platforms and search engines, as well as the increasing
de facto usage of generative AI tools as search engines.
18. To fully harness the potential benefits of AI and strengthen
democratic security, the Assembly also calls on member and observer
States of the Council of Europe to:
18.1 explore all possible uses of AI technologies to strengthen
democracy, particularly to improve the delivery of public services,
increase the protection of human rights and reduce all forms of
discrimination;
18.2 encourage and support the development of innovative ways
to engage citizens, particularly through large-scale AI-powered
consultations on difficult issues, to strengthen their involvement
in policy- and decision-making processes;
18.3 encourage and support the development of AI-powered tools
to detect, investigate and eliminate AI-generated malicious content
and cyber-attacks;
18.4 ensure the coherent and effective implementation of the
Framework Convention on Artificial Intelligence and Human Rights,
Democracy and the Rule of Law by strengthening co-ordination efforts across
signatory States.
19. To strengthen Europe’s digital sovereignty and reduce the
continent’s dependency on external providers, the Assembly calls
on member States of the Council of Europe to:
19.1 adopt robust cybersecurity measures to protect European
algorithms, models, applications, relevant networks, datasets and
critical infrastructure from malicious hackers;
19.2 focus on education, upskilling, as well as retaining and
attracting talent to create the next generation of leaders and innovators
knowledgeable about the latest technological developments and capable
of achieving sustainable socio-economic growth through them;
19.3 develop pan-European, structured “sandboxes” in which
to conduct stress tests for new technologies and tools in safe,
closed environments, under strong governance frameworks;
19.4 identify areas in which Europe still has a comparative
advantage (for example quantum computing, green technology, or the
development of specialised applications) and exploit these by allocating
sufficient financial resources and incentives, streamlining the
regulatory framework and strengthening innovation ecosystems (universities,
research centres, start-ups) and infrastructure (data centres, sovereign
clouds, high-performance computing, and chip production);
19.5 recognise that building strategic European AI technologies
and infrastructure is essential for the continent’s democratic security
and autonomy, and take concrete steps to prevent unilateral technological
dependency on foreign or private providers, while ensuring secure
and equitable access, particularly for smaller member States, public-interest
research institutions and small and medium sized enterprises;
19.6 scale up responsible innovation, through the involvement
of all stakeholders, and strengthened public-private partnerships,
in parallel with regulation;
19.7 stimulate a cultural shift towards greater tolerance of
controlled risk-taking.
20. Acknowledging the global nature of the challenge posed by
the rise of AI technologies, the Assembly calls on all States worldwide
to consider establishing a multilateral body dedicated exclusively
to overseeing AI technologies. This body could define a common language,
standards and a regulatory framework, involving all interested stakeholders,
and co-ordinate the various ongoing initiatives launched by different
multilateral organisations.
21. Furthermore, the Assembly calls on private companies to co-operate
in good faith with multilateral organisations, national and local
governments, civil society and academia, in ensuring that the development and
deployment of AI technologies, in all fields, are guided by the
respect of democratic principles, human rights and the rule of law.
22. The Assembly also encourages national parliaments to prioritise
AI in their deliberations. This should not only include legislation
and regulation, but also fostering general debates on its uses and
keeping up with relevant societal and institutional changes. One
way to achieve this could be to establish and institutionalise dedicated
parliamentary committees on AI.
23. Finally, the Assembly resolves to continue working on this
matter, through the reports prepared by its different committees
and by organising dedicated events on AI, also in relation to the
New Democratic Pact for Europe.
