알고리즘 혁신의 민사적 쟁점

If damage compensation is problematic due to hatred, discrimination, or biased results caused by algorithms, it is difficult to prove intentional or negligence, and only mental damage is likely to be problematic, so the effectiveness of relief is not great. In this regard, it is necessary to interpret the algorithm as a software product and apply the "defect estimation" provision of the 「Product Liability Act」 or apply the provisions of Article 29 of the 「Cloud Computing Development and User Protection Act」. Considering that it is remarkably difficult to prove due to information asymmetry, switching the burden of proof based on the risk liability law is considered a realistic way to remedy after death. Since it is not practically easy to receive ex-post relief for civil liability, it is also necessary to refer to the introduction of liability insurance, deduction, or compensation accumulation system, as in Article 39-9 newly established in the revision of the 「Personal Information Protection Act」. On the other hand, foreign legal trends generally erase the obligation to notify and explain algorithm service providers and give users the right to consent and choice in a way that users do not know regardless of the user's intention. In the bills actively proposed by the 21st National Assembly, we also focus on enhancing transparency and accountability by imposing an obligation to explain and agree to profiling in advance, and granting users the right to refuse or choose. However, it would be desirable to regulate similar regulations through the 「Algorithm Service User Protection Act」 in terms of legal stability or the improvement of uniform legislation, rather than confusing similar regulations.