argument: Notizie/News - Personal Data Protection Law
Source: Media@LSE
Wenlong Li and Yueming Zhang argue that regulators’ growing acceptance of legitimate interest for AI training is accompanied by fragmented data protection enforcement. Their comparative study systematically maps 19 guidelines and instruments alongside a snapshot of public enforcement actions against AI providers. The authors describe an emerging regulatory preference for legitimate interest, particularly in the private sector, while questioning the consistency of its interpretation and application. In the sampled enforcement actions, authorities often emphasised transparency and opt-out mechanisms without conclusively resolving the lawfulness of relying on legitimate interest for AI training. The authors therefore distinguish apparent agreement on a legal basis from agreement on the operational requirements that AI developers must satisfy.
Li and Zhang identify a gap between guidance emphasising assessments and enforcement concentrating on minimum safeguards, leaving uncertainty for data controllers. Their analysis also suggests that children’s data and hallucinations have received limited attention during the early stages of data protection enforcement. The authors propose making legitimate interest assessments foundational to AI training compliance and developing a minimum package of safeguards. They recommend coordinating interpretations of reasonable expectations and consent, while pooling expertise on issues where authorities diverge, including data minimisation, web scraping, accuracy and rectification. These conclusions and proposals represent the authors’ views, rather than official regulatory guidance or the position of Media@LSE or the London School of Economics and Political Science.