Artificial Intelligence Act

Generated with sparks and insights from 56 sources

Introduction

Overview: The EU Artificial Intelligence Act (AI Act) is the first comprehensive regulation on AI by a major regulator, aiming to foster safe and trustworthy AI systems while protecting fundamental rights.
Risk Categories: The Act classifies AI applications into three risk categories: unacceptable risk (banned), high-risk (subject to strict requirements), and low-risk (largely unregulated).
Banned Applications: Certain AI applications, such as social scoring and untargeted facial recognition, are banned due to their potential to infringe on citizens' rights.
High-Risk Applications: High-risk AI systems, like those used in critical infrastructure and healthcare, must meet specific legal requirements, including risk assessments and human oversight.
Transparency Requirements: General-purpose AI systems must comply with transparency requirements, including disclosing AI-generated content and ensuring compliance with EU copyright laws.
Implementation Timeline: The AI Act will enter into force 20 days after publication in the official journal, with different timelines for various provisions to become applicable.
Global Impact: The AI Act is expected to set a global standard for AI regulation, influencing international policies and practices.

Unacceptable Risk: AI applications that pose a threat to people's safety or fundamental rights are banned.
High-Risk: AI systems that significantly impact health, safety, or fundamental rights must comply with strict requirements.
Low-Risk: Applications not explicitly banned or listed as high-risk are largely unregulated.
General-Purpose AI: Added in 2023, this category includes foundation models like ChatGPT, subject to transparency requirements.
Evaluation: High-risk and general-purpose AI systems must undergo thorough evaluations before and during their lifecycle.

Social Scoring: Banned due to its potential to infringe on citizens' rights.
Untargeted Facial Recognition: Prohibited to prevent the creation of facial recognition databases.
Emotion Recognition: Banned in workplaces and schools to protect privacy.
Predictive Policing: Prohibited when based solely on profiling or assessing characteristics.
Manipulative AI: AI that manipulates human behavior or exploits vulnerabilities is banned.

Critical Infrastructure: AI systems used in critical infrastructure must meet specific legal requirements.
Healthcare: High-risk AI applications in healthcare must ensure safety and human oversight.
Education: AI systems in education and vocational training are subject to strict regulations.
Employment: AI tools used in recruitment and worker management must comply with transparency and accuracy requirements.
Law Enforcement: High-risk AI systems in law enforcement require judicial authorization and strict safeguards.

Disclosure: AI-generated content must be clearly labeled.
Copyright Compliance: General-purpose AI systems must comply with EU copyright laws.
Model Evaluations: High-impact AI models must undergo thorough evaluations.
Incident Reporting: Serious incidents involving AI systems must be reported to the European Commission.
User Awareness: Users must be informed when interacting with AI-generated content.

Entry into Force: The AI Act will enter into force 20 days after publication in the official journal.
Banned Practices: Bans on prohibited practices will apply six months after the entry into force.
Codes of Practice: These will apply nine months after the entry into force.
General-Purpose AI: Rules for general-purpose AI systems will apply 12 months after the entry into force.
High-Risk Systems: Obligations for high-risk systems will become applicable 36 months after the entry into force.

International Influence: The AI Act is expected to set a global standard for AI regulation.
Comparisons: Similar regulations are being considered in other regions, such as Brazil.
Industry Impact: International companies planning to expand to Europe will need to comply with the AI Act.
Innovation: The Act aims to boost innovation while ensuring safety and transparency.
Criticism: Some experts argue that the Act does not go far enough in banning certain AI practices.