Information about Authorised Representatives and Conformance for the upcoming EU AI Act

Introducing the EU AI Act

As the EU’s Artificial Intelligence Act general application came into force in August 2026, both EU-based and non-EU providers of certain AI systems face stringent, risk-based obligations. Central among these is the requirement for non-EU providers to designate an Authorised Representative established within the Union.

Role of the Authorised Representative

An Authorised Representative holds a written mandate from the AI provider to:

• Retain a complete copy of the system’s technical documentation for at least ten years.
• Act as the liaison with the EU AI Office and national market-surveillance authorities, providing information and documentation on request.
• Cooperate during conformity assessments, investigations or corrective-action procedures.
• Terminate the mandate and notify authorities if the provider ceases to meet its obligations.

Defining High-Risk AI vs. General-Purpose AI

• High-Risk AI Systems (Annex III): Applications where failures or biases can endanger health, safety or fundamental rights—examples include biometric identification, critical infrastructure controls and employment screening tools. These systems must undergo conformity assessments (internal or via a notified body), maintain exhaustive technical files, and implement post-market monitoring.
• General-Purpose AI Models (Article 54): Broadly capable AI models (e.g., large language or vision systems) that, even unbranded, can have systemic impact. Non-EU developers of such models must also appoint an EU Representative, publish a publicly accessible summary of training data and risk-mitigation measures, and ensure transparency around capabilities and limitations.

Key Compliance Elements

1. Technical Documentation: Detailed records covering design, training data, testing, performance metrics and risk-mitigation strategies.
2. Transparency Obligations: Clear user instructions that include the provider’s (or Representative’s) identity and contact details, system capabilities, known limitations and guidance for safe operation.
3. Conformity Assessments: Either internal quality-management checks or third-party audits, depending on risk classification.
4. Post-Market Surveillance: Ongoing monitoring to detect performance drift, emerging biases or new risks, coupled with periodic reviews and corrective measures.
5. Standards & Codes of Practice: Alignment with harmonised technical standards and sector-specific codes to demonstrate a presumption of conformity.

EU AI Act Implementation Timeline

• 1 August 2024 – The EU AI Act entered into force.
• 2 February 2025 – Prohibited AI practices and certain general provisions began to apply.
• 2 August 2025 – Obligations for providers of general-purpose AI models and certain governance provisions began to apply.
• 2 August 2026 – The majority of the AI Act’s provisions became applicable, including transparency requirements and enforcement of applicable obligations.
• 2 December 2027 – Requirements for high-risk AI systems covered by Annex III become applicable.
• 2 August 2028 – Requirements for high-risk AI systems embedded in regulated products covered by Annex I, including machinery, toys and lifts, become applicable.

This framework aims to ensure that AI systems entering the EU market are transparent, accountable and safe—backed by a designated representative who maintains regulatory readiness and supports continuous compliance.