三种欺骗
Three Forms of Deception
幻觉 Hallucination
真实案例 / Real Case
2023年，美国律师 Steven Schwartz 用 ChatGPT 写法庭文件，引用了6个根本不存在的判例。案例名称、引用编号、判决内容，全部是 AI 编造的。法官震怒，律师被罚款5000美元。AI 不知道自己在撒谎——这让它比骗子更危险。 In 2023, lawyer Steven Schwartz submitted a court brief citing 6 nonexistent cases generated by ChatGPT — complete with convincing names, citations, and holdings. All fabricated. He was fined $5,000. AI doesn't know it's lying. That makes it worse than a liar.
深度伪造 Deepfake
真实案例 / Real Case
2024年，研究人员发现数十个诈骗活动，使用CEO、新闻主播、政府官员的深度伪造视频引导受害者"投资"。受害者盯着熟悉的面孔、听着熟悉的声音，转了钱。43% 的成年网民表示在过去半年内见过深度伪造内容。 In 2024, researchers found dozens of scam campaigns using deepfake videos of CEOs, news anchors, and officials to lure victims into fake investments. 43% of adults online say they've seen a deepfake in the past six months.
虚假信息 Misinformation
真实案例 / Real Case
2024年美国大选期间，一张 AI 生成的假图——声称移民在俄亥俄州吃宠物——被前总统特朗普本人转发传播。AI 工具让一个人在几分钟内，制造出曾需数十人花数天才能完成的假新闻网站。 During the 2024 US election, an AI-generated image claiming immigrants were eating pets in Ohio was shared by Trump himself. AI tools now let one person create fake news sites in minutes that once took dozens of people days to produce.