論深偽（deepfake）不實影像之刑事管制 —以保護法益為核心

Abstract

近年來隨著科技發展，deepfake科技已經使一般人都能夠輕易地產出高度擬真的不實圖影，在帶給世人便捷與效益之餘，也同時因為不肖人士的濫用而為社會帶來了許多負面衝擊，其中包含許多與性相關之應用，然而與性無關之濫用所帶來的麻煩也不容忽視。 為了因應deepfake帶來的衝擊，我國於2023年通過刑法修正案，新增了若干針對不實影像之刑事規範。然而自修法歷程中的討論內容觀之，立法者對於不實影像可能侵害的法益為何缺乏定見，而學說上對此的討論亦莫衷一是，甚至對於修法前之既存法規是否足以因應deepfake帶來的衝擊都存在爭議。前述對於保護法益的爭論，除了與是否應另立專門之法規處理此類議題之決策有關外，尚會影響條文之構成要件應如何設計。 因此，本文旨在以探討不實影像可能涉及之保護法益為核心，進而分析出應該如何以刑法對不實影像加以規制較為適當，期待能藉此提供我國適當的立法建議。 本文首先將討論標的區分成「有混淆之虞的不實性影像使用」（類型一）、「無混淆之虞不實性影像使用」（類型二）、「有混淆之虞的與性無關之不實影像使用」（類型三）與「無混淆之虞的與性無關之不實影像使用」（類型四）等4種類型。 隨後透過文獻回顧作為方法，本文逐一分析各類型不實影像可能涉及的保護法益，並在最終發現前述之類型二與類型四基於客體欠缺足夠的欺詐性，並未侵害法益，因此不應受刑法規制；而針對類型一，對於目前主流意見所圍繞的名譽法益與隱私法益兩種見解本文皆採取反對立場，而認為應該以「形象法益」作為其保護法益；最後針對類型三，本文認為其確實為社會帶來前所未有，且既存之法規難以充分處理的新衝擊，其影響力與現行法下選擇處罰之無形偽造類型相當，因此應該修法加以因應以維護公共信用法益。 在釐清了保護法益後，本文以美國州法為主要對象，透過比較分析的方法，發現在規範客體、客觀構成要件行為、主觀意圖等若干要件設計上，各法規之間存有取捨上的歧異。本文在逐一分析所有歧異之後，以此為基礎批判了2023年度我國關於不實影像之修法存有不恰當之處，並在最終提出具體的條文設計作為修法建議。

In recent years, with the development of technology, deepfake technology has made it easy for ordinary people to create highly realistic false images. While providing convenience and benefits to society, it has also brought about many negative impacts due to the misuse by unscrupulous individuals. This includes numerous applications related to pornography, but the troubles caused by non-sexual misuse should not be ignored.

In response to the impact of deepfake technology, our country passed an amendment to the Criminal Code in 2023, which introduced several criminal regulations targeting false images. However, based on the discussions during the legislative process, it is evident that lawmakers lack a clear understanding of the legal interests that false images may violate. There is also a lack of consensus in academic discussions regarding this matter, and even the controversy surrounding whether the existing regulations prior to the amendment are sufficient to address the impact of deepfake technology. The aforementioned debates on legal interests not only relate to whether to establish separate regulations to address such issues but also to how the constituent elements of the relevant provisions should be designed.

Therefore, this article aims to focus on legal interests involved in false images and analyze how the criminal law can appropriately regulate them. The intention is to provide suitable legislative recommendations for our country.

The article first categorizes the subject of discussion into four types: "the use of sexual false images with potential confusion" (Type 1), "the use of sexual false images without potential confusion" (Type 2), "the use of non-sexual false images with potential confusion" (Type 3), and "the use of non-sexual false images without potential confusion" (Type 4).

Subsequently, through a literature review, this article analyzes the legal interests that may be involved in each type of false image. It concludes that Type 2 and Type 4, due to their lack of sufficient deception, do not violate legal interests and should not be subject to criminal regulation. Regarding Type 1, this article takes a position opposing the mainstream views surrounding reputation and privacy interests and argues that "image interests" should be the protected legal interest. Finally, concerning Type 3, this article recognizes its unprecedented impact on society and the inadequacy of existing regulations to address it adequately. Therefore, the law should be amended to protect the legal interests of public credit.

After clarifying the protected legal interests, this article primarily focuses on state laws in the United States. Through a comparative analysis, it identifies disparities in the design of several elements, such as the regulated object, objective constituent elements, and subjective intent, among different jurisdictions. Building on the analysis of these disparities, the article criticizes the 2023 amendment in our country regarding false images for its inadequacies and ultimately presents specific provisions as legislative recommendations.