A study carried out by the Director of the European Centre of Excellence on the Regulation of Robotics and Artificial Intelligence (EURA) at the request of the European Parliament’s Committee on Legal Affairs (JURI), and finalised at the end of July, analyses the question of the withdrawal of the AI Liability Directive (AILD) from the European Commission’s work programme (see EUROPE 13618/11), a decision criticised by many MEPs (see EUROPE 13620/12).
The study is based on the observation that current legislation is not adapted to the specific features of high-risk artificial intelligence systems, creating legal grey areas which, in turn, risk leading to “regulatory fragmentation” between Member States.
The critical study concludes that the proposed directive as currently drafted is too “complex” and that maintaining it as it stands would be “ineffective”.
The directive’s scope is too broad and ill-defined, with “vague” definitions, and it guarantees neither harmonised protection nor “legal simplification”, it says in detail.
The study recommends amending the directive, as the MEPs would like. It recommends refocusing its scope “on high-risk AI models, as defined by the AI Act” and introducing a system of “strict liability” for these models.
The study proposes designating a single main operator as liable: either the supplier of the AI model (its designer), or the one who deploys it (who actually uses it), or even both at the same time, in certain cases, in order to facilitate legal action.
The AILD directive is currently still being examined by the JURI committee, and amendments to the text should be voted on when Parliament resumes in September.
See the study: https://aeur.eu/f/i1t (Original version in French by Isalia Stieffatre)