EU Parliament meeting shows facial recognition still at the center of AI Act
An EU Parliament gathering – the first official debate by lawmakers about the proposed AI Act – saw European Parliament members formally agree on several administrative procedures, conformity assessments, standards and certificates.
However, according to news outlet Euractiv, most of the conversation focused on establishing the scope of facial recognition and other biometric applications, particularly in national security roles.
After viewing the initial draft of the AI Act, the European Commission called for a ban on real-time biometric identification surveillance except when the mission is identifying kidnapping victims and preventing terrorist acts.
That approach was criticized by privacy advocates, who maintain that the exceptions may pave the way for additional exceptions, eventually legitimizing biometric surveillance for what they feel are overly broad uses.
Two European Parliament officials, unnamed in Euractiv’s coverage, reportedly said heavier regulation is still a goal for most Parliament members. A majority, in fact, favor a ban on the use of facial recognition on private property.
The scope of the AI Act itself was also discussed at the meeting, with parliamentarians seeming to agree on more modest regulation.
In fact, according to Euractiv, the Parliament debated an exemption from the act for authorities in third countries and international organizations that use AI as part of international policy cooperation or as defined by judicial-cooperation agreements (if they are covered by a data adequacy decision.
Not discussed last week were proposals for a registry of AI systems believed to have the greatest potential to cause harm. Amendments on the matter, suggested over the summer, would extend compliance to anyone making a substantial modification to an algorithm.
Czech presidency wants national security opinions
On the matter of the AI Act’s scope, the Czech presidency of the EU Council recently asked whether the international convention on AI, currently being discussed by the Council of Europe, should cover matters related to national security.
Due to marked similarities between the proposed AI Act and work by the Council of Europe, the European Commission asked member states for a mandate to negotiate on behalf of the EU.
The Czech presidency recommended two options for dealing with the issue of national security. In the first scenario, the scopes of the AI Act and the international convention remain separated, with AI regulations remaining the sole responsibility of each member state. Or EU regulation and the Council of Europe’s treaty could be aligned.
“This means that questions related to national security should not be addressed by the convention contrary to what is provided for in the zero draft because national security is excluded from the scope of the AI Act,” reads a document seen by Euractiv.
Which option is chosen could determine whether the Act imposes any limitations on the use of facial recognition for purposes related to national security.
The Czech presidency also clarified that, regardless of which options EU members opt, they should follow a series of ground rules designed to support the negotiations process.
These include the EU executive having to provide a schedule of future meetings, detailing the issues to be discussed beforehand and sharing necessary information as soon as possible.
Last month, the Czech EU presidency also shared compromise text detailing new features of the proposed European digital identity infrastructure.