Yes, the EU AI Act Covers Agentic AI—Here’s How

What happened
The European Union has confirmed that its AI Act applies to agentic AI systems, including those built on large language models. The regulation requires these systems to meet transparency and risk assessment standards. This clarification comes amid claims that agentic AI falls outside the scope of the law. The European Commission has issued guidance to address these concerns and outline the obligations for providers of such AI systems.
Key details
- Agentic AI systems are covered under the EU AI Act because they meet the Act’s definition of an AI system.
- Large language models used in agentic AI are regulated as general-purpose AI models.
- Transparency requirements apply when agentic AI interacts with humans.
- Providers must conduct risk assessments, especially for models classified as posing systemic risk.
- Certain uses of AI, such as in recruitment or critical infrastructure, are pre-classified as high-risk and require comprehensive risk management.
Why it matters
Companies developing or deploying agentic AI in the EU must comply with the AI Act’s requirements, including transparency and risk management. This affects technology firms, employers, and service providers using these systems, as they face legal obligations to assess and mitigate risks. The regulation creates enforceable standards, reducing the possibility for companies to rely on self-regulation in these areas.
