Note: This article is for informational purposes only and does not constitute legal advice. (As of May 30, 2024)
Introduction
In today's rapidly evolving world of artificial intelligence (AI), the European Union (EU) has taken a proactive stance to regulate AI systems based on their potential risks. The EU AI Act aims to protect consumers and ensure the ethical development and deployment of AI technologies within the EU.
Update: As of July 2024, the AI Act is scheduled to be published in the Official Journal of the European Union, marking the formal notification of the new law. The Act will enter into force 20 days after publication.
This article provides an overview of the key aspects of the AI Act and its implications for organizations dealing with AI systems.
Who is Affected by the EU AI Act?
The EU AI Act concerns:
- Organizations producing or providing AI systems in the EU
- Organizations operating AI systems in the EU
- Consumers and regulators
Risk Categories of AI Systems
The EU has defined four risk categories to classify AI models and systems:
1. High-risk systems: Subject to strict requirements during development and deployment
2. Systems with transparency obligations: Must be marked as AI-generated
3. Unacceptable risk systems: Prohibited, with exceptions for national security or defense
4. Low or no risk systems: Remain unregulated
High-Risk AI Systems
High-risk AI systems include applications in areas such as:
- Biometrics
- Critical infrastructure
- Education and vocational training
- Employment relationships
- Aptitude assessment for essential services
- Legal decision-making
Organizations dealing with high-risk AI systems must comply with strict requirements, including:
- Establishing a risk management system
- Ensuring data governance
- Preparing technical documentation
- Implementing event logging and human oversight
- Ensuring cybersecurity measures
- Establishing a quality management system
Systems with Transparency Obligations
AI systems that interact with people, such as chatbots, must clearly disclose their AI-generated nature unless it's obvious. Operators of AI systems generating deepfakes or synthetic content must disclose that the content is artificial. AI-generated texts used to inform the public must also be marked as such, unless checked by a human before publication.
Unacceptable Risk Systems
The AI Act prohibits AI systems that:
- Manipulate people without their knowledge
- Exploit vulnerabilities
- Discriminate against individuals
Exceptions exist for national security and defense policy. Examples of prohibited systems include:
- Predictive policing
- Social scoring
- Remote identification systems without consent
- Emotion recognition
- AI used to influence vulnerable groups or manipulate behavior
Low-Risk Systems
AI systems not falling into the above categories are classified as having little to no risk and remain unregulated in the current version of the EU AI Act. However, organizations are encouraged to follow general principles of transparency, fairness, and human oversight to ensure the ethical use of these systems.
Next Steps
After the AI Act's publication and entry into force, several key milestones will follow:
- The regulation is subject to a final lawyer-linguist check and is expected to be formally adopted before the end of the current legislative session.
- It will become fully applicable 24 months after its entry into force, with specific provisions becoming applicable at different intervals.
This phased approach allows organizations and regulatory bodies time to prepare for full compliance with the new regulations.
Summary
The EU AI Act sets a global precedent in regulating artificial intelligence based on risk levels. Organizations developing, providing, or operating AI systems in the EU must assess their systems' risk categories and comply with the corresponding requirements to ensure the safe and ethical use of AI technologies.
By staying informed about the EU AI Act and its implications, organizations can navigate the evolving landscape of AI regulation and contribute to a future where artificial intelligence benefits society while minimizing potential risks.
Sources
EU Artificial Intelligence Act | Up-to-date developments and analyses of the EU AI Act
.png)