AI law: MEPs submit thousands of amendments before the start of negotiations

Each political group in the European Parliament submitted a few hundred amendments to the future AI law, thus setting the tone for the discussions to come.

These amendments, thousands in total, are the prelude to complex negotiations which should begin before the summer, during which the co-rapporteurs Brando Benifei for the Internal Market Committee (IMCO) and Dragoș Tudorache for the Civil Liberties Committee (LIBE ) will seek to secure a majority through compromises.

According to an official of the European Parliament, the deputies of the two main committees are almost evenly distributed around the centre-right and centre-left axes, and on the most controversial points it will be a “calculation game”.


One of the most controversial topics is the very definition of artificial intelligence (AI).

MEP Mr Benifei proposes a broader definition as well as deleting the list of AI techniques and approaches in Annex I so that the regulation is future proof.

In contrast, the centre-right European People’s Party (EPP) insists on the definition agreed at OECD level. PPE also presented a definition of machine learning as the ability to detect trends without being explicitly programmed for a specific task.

Both co-rapporteurs maintain their opposition to limiting an exception for AI as a general-purpose technology. Furthermore, the EPP has proposed to lighten the burden of the obligations incumbent on these suppliers, by introducing different requirements for new, old and original suppliers.


Liberal MEP Mr Tudorache presented a new article for AI applications in the metaverse to be included in the scope of the regulation, including a reference to blockchain-based currencies and NFTs. He also proposed applying the rules to suppliers that are not located or operating in the EU under certain circumstances.

Prohibited practices

Green MPs have made major proposals in terms of prohibited practices, expanding this category to include biometric categorization, emotion recognition and any automated tracking of human behavior. These practices include recommendation systems that consistently offer misinformation and illegal content, uses in law enforcement, migration, labor and education.

Biometric identification and recognition

In a major change of position, Mr Tudorache joined the Social Democrats and the Greens in arguing for a total ban on biometric recognition, eliminating the exceptions included in the initial proposal.

With regard to biometric identification, Mr Tudorache and the EPP both propose to exclude consented authentication to access a service, device or location from the definition.

The Greens group has presented a ban on personal biometric databases based on the extraction of information from the Internet, such as that of the controversial company Clearview AI.

High risk systems

The centre-right has proposed a new regime changing the automatic classification of systems falling under the list of sectors mentioned in Annex III into a list of “critical use cases”.

Based on these uses, AI providers should self-assess whether their systems pose significant risks to health, safety and fundamental rights.

In addition, the EPP proposed that requirements for high-risk applications could be partially or fully removed if, by fulfilling the regulation’s risk mitigation measures, systems have sufficiently mitigated the risk through operational countermeasures. or built-in devices.

Conservative lawmakers also excluded systems designed to assess creditworthiness from the high-risk list. In contrast, Green MPs extended this category to media recommendation systems, algorithms used in health insurance processes, payments, and debt collection.

The Greens group has established much stricter environmental requirements and strengthened the role of third parties and notified bodies. In addition, fundamental rights impact assessments have been made mandatory for all vendors and deployers.

Legal clarity

The EPP proposed the creation of a new article defining reliability by presenting a series of principles on how the concept should be applied to AI in terms of technical requirements and standards.

Additionally, in order to better define how the concept of accuracy, reliability, robustness and cybersecurity should be applied, conservative legislators have introduced provisions requiring national metrology and calibration authorities to provide guidance not binding.


The EPP followed the proposal of the French Presidency of the Council of the EU (PFUE) to involve stakeholders in the work of the European Council on Artificial Intelligence and proposed to give this body more autonomy in the setting its own agenda.

The Greens want the European Data Protection Supervisor (EDPS) to provide the Council secretariat. The EDPS would act as a supervisory authority for large companies whose breaches have a “community dimension”in accordance with the criteria of the European Merger Regulation.

Lawmakers across the political spectrum appear to be in favor of giving the regulator more investigative power, with the Greens particularly ambitious when it comes to corrective action.

Advertising and dark patterns

A separate amendment was presented by MEP Maria-Manuel Leitão-Marques, with the support of the Tracking Free Ad Coalitiona coalition of political leaders, civil society organizations and businesses across the EU who have pledged to end ad tracking on the internet, to include AI systems used to serve advertising online in the list of high risk systems.

The Greens group also added a paragraph stating that the requirement for transparency on automated decision-making should help address power imbalances in digital environments and tackle dark patterns.


Liberal and Conservative MEPs propose a general reduction in fines, with the EPP notably including an exception for SMEs and adding factors to be taken into account such as intent, negligence and cooperation in the calculation of the fine by the authorities .

By contrast, centre-left MEP Mr Benifei is pushing for a general increase in sanctions and for size and market share to be removed from the criteria authorities should use to impose a sanction.

We would like to thank the author of this short article for this outstanding content

AI law: MEPs submit thousands of amendments before the start of negotiations

Find here our social media profiles and other pages related to it.