The European Parliament has endorsed new rules for Artificial Intelligence (AI), marking a significant step towards ethical AI development. These rules include bans on biometric surveillance, emotion recognition, and predictive policing AI systems, alongside tailored regulations for general-purpose AI and foundation models like GPT. The legislation operates on a risk-based approach, prohibiting AI practices that pose unacceptable risks to safety and privacy, and expanding high-risk classifications to include AI influencing political campaigns and social media. Transparency measures for AI providers are mandated, emphasizing the protection of rights and the environment. The new law aims to balance innovation support with citizens’ rights, allowing for complaints about AI decisions. Co-rapporteurs emphasized the importance of this legislation as a global benchmark for human-centric AI governance. The draft must be approved by the full Parliament before final negotiations with the Council.
| name | description | change | 10-year | driving-force | relevancy |
|---|---|---|---|---|---|
| First Global AI Regulations | Europe is drafting the world’s first comprehensive AI regulations. | Shifting from unregulated AI development to structured, ethical guidelines. | Global AI governance frameworks may emerge, inspired by Europe’s model. | Increasing public concern over AI safety and ethical implications. | 5 |
| Ban on Biometric Surveillance | Proposed bans on various intrusive AI practices, including biometric surveillance. | Transitioning from widespread surveillance to regulated, ethical practices. | Potential decline in invasive surveillance practices globally, with privacy protections. | Demand for civil liberties and human rights protections. | 4 |
| Transparency in AI Systems | New rules require transparency and accountability for AI systems. | From opaque AI systems to clearer guidelines and accountability measures. | AI systems may become more transparent, fostering user trust and safety. | Public demand for accountability in technology. | 4 |
| High-Risk AI Classification | Expansion of high-risk AI areas to include health and political influence. | From vague classifications to specific, regulated high-risk categories. | Improved safety and ethical standards for high-risk AI applications. | Need to protect fundamental rights and public safety. | 4 |
| Support for Innovation with Regulations | Regulatory sandboxes and exemptions for research promote AI innovation. | From restrictive regulations to supportive frameworks for AI development. | A thriving innovation ecosystem could emerge, balancing ethics and technology. | Desire to foster innovation while ensuring ethical standards. | 3 |
| Empowerment of Citizens’ Rights | Citizens gain the right to complain and receive transparency about AI decisions. | From passive users to empowered individuals with rights in AI governance. | Citizens may have a stronger voice in AI governance and accountability. | Growing awareness of individual rights in technology usage. | 4 |
| Global Leadership in AI Legislation | EU aims to set a global standard for AI legislation. | From fragmented regulations to unified, comprehensive global standards. | The EU could be seen as a model for global AI governance frameworks. | Desire for regulatory coherence and international leadership in technology. | 5 |
| name | description | relevancy |
|---|---|---|
| Biometric Surveillance Risks | The potential for misuse of biometric data leading to privacy violations and discrimination in public spaces. | 5 |
| Manipulative AI Practices | AI systems that utilize subliminal techniques or exploit vulnerabilities might cause significant harm to individuals’ autonomy and decision-making. | 5 |
| Social Scoring Systems | Using AI for social scoring could lead to unjust classifications and inequality, impacting individuals’ rights and freedoms. | 5 |
| High-Risk AI Deployments | AI influencing political campaigns or voter behavior poses risks to democratic processes and public trust. | 4 |
| Transparency in AI Systems | The need for transparency in generative AI to prevent the spread of illegal or harmful content is critical for public safety and trust. | 4 |
| Environmental Impact of AI | The growing environmental impact of AI technologies necessitates stringent regulations to ensure sustainability. | 4 |
| Discriminatory AI Practices in Law Enforcement | The use of emotion recognition and predictive policing can perpetuate bias and discrimination within law enforcement. | 5 |
| Public Complaints about AI Decisions | Guaranteeing citizens’ rights to complain about AI decisions is essential for accountability and ethical governance. | 4 |
| Controlled Testing of AI | Regulatory sandboxes may inadvertently lead to inadequate oversight if not properly monitored, risking harmful AI deployment. | 3 |
| name | description | relevancy |
|---|---|---|
| Human-centric AI Governance | Establishment of rules ensuring AI systems are overseen by people and prioritize human rights and ethics. | 5 |
| Transparency in AI Systems | Requirements for AI systems to be transparent and traceable to build public trust and accountability. | 5 |
| Risk-based Regulation | Implementation of a risk-based approach to classify AI systems and establish obligations based on their risk level. | 4 |
| Prohibition of Harmful AI Practices | Bans on AI practices that pose unacceptable risks to safety or privacy, such as biometric surveillance and predictive policing. | 5 |
| Promotion of Innovation with Safeguards | Incentives for AI innovation through exemptions for research and open-source projects while ensuring citizen protections. | 4 |
| Strengthened Citizens’ Rights | Empowerment of citizens to file complaints and receive explanations regarding high-risk AI systems affecting their rights. | 4 |
| Global Leadership in AI Regulation | EU positioning itself as a leader in establishing comprehensive AI regulations that can guide global practices. | 5 |
| name | description | relevancy |
|---|---|---|
| Artificial Intelligence (AI) Regulation | Legislation aiming to create rules for AI development, ensuring safety, transparency, and ethical standards in Europe. | 5 |
| Foundation Models | Fast-evolving AI models like GPT that require compliance with transparency and risk management rules. | 4 |
| Generative AI | AI systems that generate content, needing disclosure of AI-generated content and compliance with legal standards. | 4 |
| Regulatory Sandboxes | Controlled environments established by public authorities to test AI before deployment, fostering innovation while ensuring safety. | 4 |
| name | description | relevancy |
|---|---|---|
| Regulation of AI Technology | The emergence of the world’s first comprehensive rules on AI, focusing on ethical development and oversight. | 5 |
| Biometric Surveillance Concerns | Growing concerns over the use of biometric surveillance technologies and their implications for privacy and human rights. | 5 |
| High-Risk AI Classification | The classification of AI systems into high-risk categories highlights the potential dangers associated with AI applications. | 4 |
| Transparency in AI Systems | The push for transparency measures in AI, particularly in generative models, to maintain trust and accountability. | 4 |
| Citizens’ Rights in AI Decision-Making | Strengthening citizens’ rights to complain about AI systems and receive explanations for decisions that affect them. | 4 |
| Innovation vs Regulation Balance | Finding a balance between regulating AI and promoting innovation, particularly for startups and SMEs. | 4 |
| Global Leadership in AI Regulation | The EU’s role in setting a global standard for AI regulation may influence international policies and practices. | 3 |
| Environmental Impact of AI | Incorporating environmental considerations into AI regulations may become increasingly important as technology advances. | 3 |