Sora 2数手指翻车，奥特曼成第一批“受害者”，被AI玩成最惨打工人

Core Points - The article discusses the challenges faced by AI video generation models, particularly Sora 2 and Veo3, in accurately counting fingers while verbally counting numbers [1][2][6] - OpenAI's new social media application, Sora, has generated mixed reactions, with some users enjoying the quirky AI-generated content featuring Ultraman, while others express concerns about the implications of deepfake technology [8][11][13] - OpenAI researchers have voiced their concerns regarding the ethical implications of releasing Sora, balancing the excitement of technological advancements with the potential risks associated with deepfake videos [13][14][15] Group 1: AI Video Generation Challenges - Sora 2 and Veo3 failed to accurately count fingers while verbally counting, even after refining the prompt used for testing [1][2][6] - Users have created various tests for the AI models, including simulating water pouring into a glass, which Sora 2 performed well but still had minor flaws [6] Group 2: User Reactions to Sora - The AI-generated videos featuring Ultraman have led to a mix of amusement and concern among users, with some finding the content entertaining while others worry about the implications of deepfake technology [8][10][11] - Ultraman's portrayal in these videos has sparked discussions about the potential misuse of AI-generated content and copyright issues [11] Group 3: OpenAI Researchers' Concerns - OpenAI researchers have expressed their worries about the ethical implications of Sora's release, emphasizing the need for responsible AI development [13][14] - The balance between showcasing innovative technology and adhering to a non-profit mission has been a topic of debate among OpenAI staff [15][16]