The EU’s AI Act is Now in Force

The European Union’s risk-based regulation for applications of artificial intelligence has come into effect starting today.

This landmark legislation introduces a series of staggered compliance deadlines that will impact various types of AI developers and applications, with full compliance required by mid-2026.

Initial Compliance Deadlines

The first key deadline is just six months away. This will enforce bans on specific AI uses, such as law enforcement’s use of remote biometrics in public places. It’s a significant step toward regulating AI in high-stakes areas.

Most AI applications, which are considered low or no risk, will be outside the scope of this regulation. However, high-risk AI applications like biometrics and facial recognition will require registration in an EU database and adherence to strict risk and quality management obligations.

High-Risk AI Applications

High-risk AI systems used in areas like education and employment will need to meet stringent requirements. Developers must ensure these systems comply with risk and quality management standards to protect users.

This classification aims to safeguard sensitive domains where AI can significantly impact people’s lives and opportunities.

Limited Risk AI Technologies

Limited risk AI technologies, such as chatbots, aren’t off the hook. They must meet transparency requirements to prevent user deception. Adding a simple disclaimer at the beginning of your chatbot conversation should be enough.

Users should be aware they are interacting with AI, ensuring a level of trust and clarity in AI-driven interactions.

The law’s focus on transparency is a preventive measure against misuse and unintended consequences of AI technology.

General Purpose AI (GPAI)

Developers of general-purpose AIs (GPAIs) will generally face light transparency requirements. However, the most powerful models will need to conduct risk assessments and implement mitigation measures.

The exact compliance steps for GPAI developers are still under discussion. Codes of Practice are expected to be finalized by April 2025.

Consultation and Codes of Practice

The AI Office has initiated a consultation to involve stakeholders in developing these Codes of Practice. This collaborative approach aims to create balanced and effective regulations.

OpenAI, known for its GPT models, has expressed its intention to work closely with the EU AI Office, providing technical documentation and guidance for compliance.

Guidance for AI Developers

OpenAI’s primer suggests that organizations start by classifying their AI systems to determine their compliance obligations. Identifying whether they are providers or deployers is crucial for understanding their responsibilities.

Consulting with legal counsel can help navigate the complex issues surrounding AI regulation and ensure adherence to the new rules.

Implications for AI Developers

AI developers must now navigate a new regulatory landscape. The EU’s risk-based approach means that only the most significant applications face stringent oversight.

This regulation could serve as a model for other countries, potentially leading to global standards for AI development and usage.

The EU’s AI Act marks a pivotal moment in AI regulation, setting a precedent for risk-based approaches to technology oversight.

As developers adapt to these new rules, the AI landscape is set to change dramatically, emphasizing transparency, risk management, and ethical considerations.

About The Author

Emmanuel Kesse

See author's posts