On 25 November 2021, the German Bar Association (Deutscher Anwaltverein - DAV) published a position paper on the Proposal of the European Commission for a Regulation of the European Parliament and of the Council Laying Down Harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and Amending Certain Union Legislative Acts, adopted on 21 April 2021 (COM (2021) 206 final) (on the Commission’s Proposal → eucrim 2/2021, 77). The DAV welcomed the risk-based approach of the AIA Proposal but points out that the binary classification into high-risk/non-high-risk AI systems leaves less room for differentiation of other risk levels.The association welcomed the EU Commission’s proposal but criticised the definition of AI systems in Art. 3(1) of the proposal as being too narrow with regard to future developments, as it refers to human-defined objectives. It does not object to Art. 5 of the proposal’s lists of intolerable and prohibited AI systems. However, the DAV advocates clear criteria that would help distinguish between AI systems that are prohibited and AI systems that are permitted.

The DAV further welcomed the prohibition of social scoring but expressed regret that the prohibition has been softened by the conditions mentioned in (i) and (ii). It sees room for broad interpretation in the wording of the conditions (e.g. terms such as “unrelated to the contexts” (i) or “unjustified or disproportionate” (ii)).

The association criticizes that only biometric identifications in real time-situations have been banned and that Art. 5 of the proposal only prohibits biometric identification systems when used for law enforcement purposes. The DAV regrets the absence of a general ban on AI systems that take independent judicial decisions, a “predictive policing” ban, and a complete ban of polygraphs. It joins the CCBE Position Paper in its call for a ban on AI systems in the areas of migration, asylum, and border control management until they have been independently assessed for compliance with international human rights standards.

With regard to use of AI in the area of law enforcement, the DAV stressed that, in cases in which decisions are based on data or results are produced by an AI system, the parties and/or their lawyers must be able to access this AI system in order to assess its characteristics, the data used, and the relevance of the results it provides. For the use of AI in the justice area, the association recommends that detailed principles and guidelines be established and that AI systems only be introduced if sufficient safeguards against discrimination and bias are in place.