The EU AI Act Is Now Real
The EU Artificial Intelligence Act entered into force on 1 August 2024, with obligations being phased in over three years. As of February 2025, prohibited AI practices are already banned. By August 2026, all high-risk AI system requirements will be fully enforceable.
If your business uses AI tools, this regulation affects you — whether you're developing AI or simply deploying it.
Timeline of Key Dates
Who Does This Affect?
The EU AI Act applies to:
If your European business uses ChatGPT, Copilot, or any AI tool in operations that affect people's rights or safety, you are a "deployer" under the Act.
Risk Classification
The Act classifies AI systems into four risk categories:
Unacceptable Risk (Banned)
High Risk (Strict Obligations)
Limited Risk (Transparency Obligations)
Minimal Risk (No Obligations)
What Deployers Must Do
If you're using AI tools classified as high-risk:
Impact on AI Tool Selection
The EU AI Act creates new criteria for evaluating AI tools:
Prefer tools that support compliance documentation
Tools like Holistic AI and Vanta provide automated EU AI Act risk classification and evidence collection.
Prefer tools with ISO 42001 certification
ISO/IEC 42001 is the international standard for AI management systems. Tools with this certification (Claude, Augment Code) have formally implemented AI governance frameworks.
Prefer EU-headquartered providers for high-risk use cases
For AI systems in HR, healthcare, finance, or legal — where the Act imposes the strictest obligations — using EU-headquartered providers (DeepL, Mistral AI, Celonis) simplifies compliance by keeping the entire data processing chain under EU jurisdiction.
Audit your current AI tools
Map every AI tool in your organisation against the risk classification framework. You may find that some existing tools are being used in high-risk contexts without appropriate governance.
Tools to Help You Comply
| Tool | What It Does | HQ |
|---|---|---|
| Holistic AI | AI governance, bias detection, EU AI Act readiness | London, UK |
| Vanta | Compliance automation with EU AI Act module | San Francisco, US |
| OneTrust | Privacy and AI governance platform | Atlanta, US |
| 2021.AI (GRACE) | On-premise AI with built-in governance | Copenhagen, DK |
Next Steps
The EU AI Act is complex, but its core message is simple: if you use AI in ways that affect people's lives, you must do so responsibly and transparently. TrustKit's compliance scores help you start by choosing the right tools.
