This in-depth analysis (IDA) explores the most prominent actors, cases and techniques of algorithmic authoritarianism together with the legal, regulatory and diplomatic framework related to AI-based biases as well as deliberate misuses. With the world leaning heavily towards digital transformation, AI’s use in policy, economic and social decision-making has introduced alarming trends in repressive and authoritarian agendas. Such misuse grows ever more relevant to the European Parliament, resonating with its commitment to safeguarding human rights in the context of digital trans-formation. By shedding light on global patterns and rapidly developing technologies of algorithmic authoritarianism, this IDA aims to produce a wider understanding of the complex policy, regulatory and diplomatic challenges at the intersection of technology, democracy and human rights. Insights into AI’s role in bolstering authoritarian tactics offer a foundation for Parliament’s advocacy and policy interventions, underscoring the urgency for a robust international framework to regulate the use of AI, whilst ensuring that technological progress does not weaken fundamental freedoms. Detailed case studies and policy recommendations serve as a strategic resource for Parliament’s initiatives: they highlight the need for vigilance and proactive measures by combining partnerships (technical assistance), industrial thriving (AI Act), influence (regulatory convergence) and strength (sanctions, export controls) to develop strategic policy approaches for countering algorithmic control encroachments.
Authors
Directorate-General for External Policies of the Union, European Parliament, ünver, Akin
Related Organizations
- Catalogue number
- QA-02-24-534-EN-N QA-02-24-534-EN-C
- Citation
- European Parliament, Directorate-General for External Policies of the Union, Ünver, A., Artificial intelligence (AI) and human rights – Using AI as a weapon of repression and its impact on human rights – In-depth analysis , Publications Office of the European Union, 2024, https://data.europa.eu/doi/10.2861/52162
- DOI
- https://data.europa.eu/doi/10.2861/907329 https://data.europa.eu/doi/10.2861/52162
- ISBN
- 978-92-848-1855-6
- Pages
- 119
- Published in
- Belgium
- Themes
- Fundamental rights , Information technology and telecommunications
Table of Contents
- Artificial Intelligence_and_HR.pdf 1
- List of abbreviations 7
- 1 Introduction 9
- 2 Defining repression 10
- 2.1 Scientific definitions 10
- 2.2 Policy definitions 13
- 3 Case studies on algorithmic authoritarianism and surveillance 15
- 3.1 Case and time selection criteria 15
- 3.2 Advanced algorithmic control at scale: The case of China 17
- 3.2.1 Xinjiang region and the persecution of Uighur minority 17
- 3.2.2 China’s social credit system: A technical examination 20
- 3.2.3 Key actors and entities involved in Chinese algorithmic authoritarianism 21
- 3.2.4 Transnational involvement of China’s AI policy and misuse 23
- 3.3 Russian algorithmic authoritarianism: The Yarovaya Law of 2016 and evolution after the 2022 Ukraine invasion 25
- 3.3.1 Evolution and current state after Russia’s 2022 war on Ukraine 27
- 3.3.2 Russian state actors and algorithmic authoritarianism practices 30
- 3.4 Iranian AI-based repression systems: Silencing dissent and suppressing opposition 33
- 3.4.1 Key actors in Iranian algorithmic authoritarianism 36
- 3.4.2 Transnational surveillance and control 38
- 3.5 Egypt or the quest to prevent another Tahrir 39
- 3.6 Algorithmic authoritarianism in Sub-Saharan Africa: The case of Ethiopia and beyond 41
- 4 The role of democracies: Algorithmic bias and technology exports 43
- 4.1 US AI-based systems: Concerns over surveillance and privacy 46
- 4.1.1 Private companies in US domestic AI-based monitoring systems and American AI Systems Exports 50
- 4.1.2 Legislative and judicial check in the USA 51
- 4.2 European high-technology exports 52
- 5 Assessing the effectiveness of the current international regulatory framework and governance initiatives on AI 55
- 5.1 The EU 55
- 5.1.1 The AI Act 57
- 5.1.2 The Ethics Guidelines for Trustworthy AI 58
- 5.1.3 Other EU initiatives 59
- 5.2 The Council of Europe 60
- 5.3 Non-binding international initiatives 61
- 5.3.1 The United Nations and the UNESCO Guideline Evaluation 61
- 5.3.2 The OECD, non-binding AI principles and the Global Partnership on AI 63
- 5.3.3 Expert forums 66
- 5.4 State-led initiatives outside the EU 68
- 5.4.1 The USA 68
- 5.4.2 China 69
- 5.4.3 India 71
- 6 Key recommendations 72
- 6.1 Recommendations for the EU 72
- 6.2 Recommendations for the EP 78
- 6.3 Final conclusions 79
- 7 References 81
- 8 Annexes 97
- 8.1 Techniques, tactics and procedures of algorithmic authoritarianism and bias: An overview of technical repertoires 97
- 8.1.1 Automated Content Filtering (ACF) 97
- 8.1.2 Sentiment analysis 99
- 8.1.2.1 Technical foundations of automated sentiment analysis 99
- 8.1.2.2 Impacts on freedom to access information 100
- 8.1.3 Deep packet inspection 100
- 8.1.4 Facial recognition and surveillance 101
- 8.1.4.1 Technical foundations of facial recognition in surveillance 101
- 8.1.4.2 Applications and impacts on repression 102
- 8.1.5 Predictive policing 103
- 8.1.5.1 Technical foundations of predictive policing 104
- 8.1.5.2 Applications and repression implications 104
- 8.1.6 Deepfake technology 106
- 8.1.6.1 Technical foundations of deepfakes 106
- 8.1.6.2 Applications and repression implications 107
- 8.1.7 Gait detection 108
- 8.1.7.1 Applications and repression implications 109
- 8.2 Current trends in AI abuse for repression 110
- 8.2.1 Outcomes and motivations: Why do governments engage in algorithmic authoritarianism? 110
- 8.2.2 Not all algorithmic authoritarianism plans succeed: Intended vs real effects of AI authoritarianism 111
- 8.2.2.1 User adaptation and resistance 111
- 8.2.2.2 Technological evasion 112
- 8.2.2.3 Algorithmic inefficiency 113
- 8.2.2.4 The over-reliance on technological solutions 114
- 8.2.3 Impact of AI technologies on freedoms and rights 115
- 8.2.3.1 Political Imprisonment 116
- 8.2.3.2 Torture 116
- 8.2.3.3 Killings 117
