Inferensys

Glossary

Market Withdrawal Notification

The formal regulatory obligation under the EU AI Act requiring providers to immediately inform the market surveillance authority and update the EU database when a registered high-risk AI system is permanently recalled or withdrawn from the Union market.
Risk analyst performing AI risk assessment on laptop, risk matrices visible, casual office risk session.
REGULATORY OBLIGATION

What is Market Withdrawal Notification?

A formal regulatory duty requiring AI system providers to immediately inform authorities when a registered system is recalled or permanently removed from the EU market.

A Market Withdrawal Notification is the legally mandated process by which a provider of a registered high-risk AI system must formally inform the relevant market surveillance authority and update the EU AI Act database when that system is permanently recalled or withdrawn from the Union market. This obligation ensures the public registry accurately reflects the current compliance status and availability of all regulated systems, preventing reliance on outdated or decommissioned entries.

The notification must include the Unique Registration ID, the rationale for withdrawal—whether due to safety concerns, non-compliance findings, or commercial decisions—and the effective date of removal. This process is distinct from a Registration Suspension imposed by a National Competent Authority; it is a proactive duty of the provider to maintain the integrity of the regulatory database and trigger any necessary downstream Incident Reporting Linkage if the withdrawal is related to a serious risk.

MARKET WITHDRAWAL COMPLIANCE

Frequently Asked Questions

Clarifying the regulatory obligations and technical procedures required when a high-risk AI system must be recalled or permanently removed from the European Union market.

A Market Withdrawal Notification is the formal, legally mandated communication from an AI system provider to the relevant market surveillance authority and the EU AI Act Database confirming that a specific high-risk AI system has been permanently removed from the Union market. This notification is not merely an administrative update; it is a critical compliance control that triggers the deactivation of the system's Unique Registration ID in the public-facing database. The obligation ensures that non-compliant, unsafe, or obsolete algorithmic systems are no longer available for deployment or sale, thereby maintaining the integrity of the regulatory ecosystem. The notification must detail the reasons for withdrawal—such as a failed post-market monitoring audit, a discovered substantial modification without re-assessment, or a voluntary strategic recall—and confirm that all distributors and importers have been instructed to cease distribution immediately.

Regulatory Recall Protocol

Core Components of a Market Withdrawal Notification

The formal, non-negotiable obligation to inform the market surveillance authority and update the EU database when a registered AI system is recalled or withdrawn from the market.

01

Immediate Authority Notification

The provider must immediately inform the relevant National Competent Authority (NCA) of the member state where the system was registered. This notification cannot be delayed and must precede or coincide with the public announcement of the withdrawal. The alert triggers an official review to determine if the withdrawal is due to a serious incident or non-compliance that could affect other systems.

< 15 days
Max Notification Window
02

EU Database Status Update

The provider is legally required to update the system's status in the EU AI Act Database. The Unique Registration ID must be flagged as 'withdrawn' or 'recalled.' This action ensures that the Digital Product Passport and all linked compliance documentation reflect the inactive market status, preventing downstream users or importers from inadvertently deploying a non-compliant system.

03

Reasoning and Corrective Actions

The notification must include a detailed explanation of the withdrawal's root cause. This dossier should specify:

  • Non-compliance nature: Whether the withdrawal is voluntary or mandated by an authority.
  • Risk assessment: An analysis of the risk posed to health, safety, or fundamental rights by the system's continued operation.
  • Remediation plan: A clear outline of the corrective measures being taken, such as a software patch, hardware retrofit, or permanent decommissioning.
04

End-User Communication Protocol

Providers must proactively communicate the withdrawal to all known deployers and end-users. This communication must be clear, transparent, and include the Residual Risk Disclosure that prompted the action. The goal is to ensure that all operators cease use of the non-compliant system immediately, mitigating potential harm while the corrective action is implemented.

05

Supply Chain Traceability

The withdrawal notification triggers a cascade of obligations across the supply chain:

  • Importers must verify that the non-EU manufacturer has executed the withdrawal and halt further placements.
  • Distributors must quarantine remaining stock and cooperate with the NCA to trace deployed units.
  • Authorized Representatives serve as the primary liaison for non-EU providers, ensuring the notification reaches the correct Union authorities.
06

Incident Reporting Linkage

If the market withdrawal is a direct consequence of a serious incident, the notification must be formally linked to the Incident Reporting Portal. This creates an auditable chain of evidence connecting the system's failure mode to the decision to remove it from the market. This linkage is critical for the NCA's post-market monitoring and potential enforcement actions.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.