LISTEN NOW: Evolution of Data Controls in the Era of Generative AI

View

Article 46: Derogation From Conformity Assessment Procedure | EU AI Act

Contributors

Anas Baig

Product Marketing Manager at Securiti

Syed Tatheer Kazmi

Associate Data Privacy Analyst, Securiti

CIPP/Europe

Article 46 of the AI Act provides information related to derogation from conformity assessment procedure.

A market surveillance authority can authorize a high-risk AI system to be placed on the market or into service within the territory of the Member State concerned, despite non-compliance with Article 43, if there are compelling reasons related to public security, human life and health, environmental protection, or protection of key industrial and infrastructural assets.

Such authorization will only be for a limited period while the relevant conformity assessment procedures are carried out, considering exceptional reasons that justify the derogation. These procedures must be completed without unnecessary delay.

In cases of urgency due to exceptional reasons of public security or in the case of a specific, substantial, and imminent threat to the life or physical safety of natural persons, the law enforcement authorities or civil protection authorities may put a high-risk AI system into service without appropriate authorization as described above if such authorization is requested during or after the use of such a system without undue delay. In case the request is rejected, the use of such a system must be ceased immediately with the outputs and results discarded in a similarly immediate manner.

The authorization can only be issued if the market surveillance authority concludes that the high-risk AI system properly complies with the above-outlined requirements. The market surveillance authority must duly inform the Commission and other Member States of any authorizations it issues. However, this obligation does not cover sensitive operational data related to the activities of law enforcement authorities.

If a Member State or the Commission does not object within 15 days of receiving any information related to such an authorization by the market surveillance authority of a member state, the authorization will be deemed justified.

However, if objections are raised within this timeframe by a member state or the Commission against an authorization issued by a market surveillance authority of another Member State, the Commission must consult with the relevant Member State. The concerned operators will be consulted and given the chance to present their views. After considering these inputs, the Commission must determine whether the authorization is justified, with the decision duly communicated to both the operators and the Member State.

If the Commission finds the authorization unjustified, the market surveillance authority of the concerned Member State will withdraw it.

Suscríbase a nuestro boletín

Obtenga toda la información más reciente, actualizaciones de leyes y más en su bandeja de entrada


Compartir

Videos

Spotlight Talks

Spotlight 1:10:56

Cómo Prepararse y Actuar Positivamente ante la Ley 81

Cómo Prepararse yActuar Positivamente ante la Ley 81
Ver ahora View

Latest

View More

From Trial to Trusted: Securely Scaling Microsoft Copilot in the Enterprise

AI copilots and agents embedded in SaaS are rapidly reshaping how enterprises work. Business leaders and IT teams see them as a gateway to...

The ROI of Safe Enterprise AI View More

The ROI of Safe Enterprise AI: A Business Leader’s Guide

The fundamental truth of today’s competitive landscape is that businesses harnessing data through AI will outperform those that don’t. Especially with 90% of enterprise...

Data Security Governance View More

Data Security Governance: Key Principles and Best Practices for Protection

Learn about Data Security Governance, its importance in protecting sensitive data, ensuring compliance, and managing risks. Best practices for securing data.

AI TRiSM View More

What is AI TRiSM and Why It’s Essential in the Era of GenAI

The launch of ChatGPT in late 2022 was a watershed moment for AI, introducing the world to the possibilities of GenAI. After OpenAI made...

Managing Privacy Risks in Large Language Models (LLMs) View More

Managing Privacy Risks in Large Language Models (LLMs)

Download the whitepaper to learn how to manage privacy risks in large language models (LLMs). Gain comprehensive insights to avoid violations.

View More

Top 10 Privacy Milestones That Defined 2024

Discover the top 10 privacy milestones that defined 2024. Learn how privacy evolved in 2024, including key legislations enacted, data breaches, and AI milestones.

Comparison of RoPA Field Requirements Across Jurisdictions View More

Comparison of RoPA Field Requirements Across Jurisdictions

Download the infographic to compare Records of Processing Activities (RoPA) field requirements across jurisdictions. Learn its importance, penalties, and how to navigate RoPA.

Navigating Kenya’s Data Protection Act View More

Navigating Kenya’s Data Protection Act: What Organizations Need To Know

Download the infographic to discover key details about navigating Kenya’s Data Protection Act and simplify your compliance journey.

Gencore AI and Amazon Bedrock View More

Building Enterprise-Grade AI with Gencore AI and Amazon Bedrock

Learn how to build secure enterprise AI copilots with Amazon Bedrock models, protect AI interactions with LLM Firewalls, and apply OWASP Top 10 LLM...

DSPM Vendor Due Diligence View More

DSPM Vendor Due Diligence

DSPM’s Buyer Guide ebook is designed to help CISOs and their teams ask the right questions and consider the right capabilities when looking for...

What's
New