📄 中文摘要
2026年2月27日，安全研究员Taimur Khan在Lovable平台上发现了一款应用程序存在16个安全漏洞，其中6个为关键漏洞。这款应用是一个AI驱动的教育科技工具，旨在生成AI考试和评分学生提交的作业。漏洞导致加州大学伯克利分校、加州大学戴维斯分校及K-12学校的学生数据泄露，影响了约18,000名用户。此事件揭示了在AI应用开发中，安全性的重要性和潜在风险，提醒开发者在构建AI工具时必须重视安全防护措施。
可爱的应用程序暴露了18,000用户：未曾预见的Vibe编码安全危机
📄 English Summary
Lovable App Exposes 18,000 Users: The Vibe Coding Security Crisis Nobody Saw Coming
On February 27, 2026, security researcher Taimur Khan discovered 16 vulnerabilities, including 6 critical ones, in a single app hosted on Lovable's platform. This AI-powered EdTech tool, designed to generate AI exams and grade student submissions, leaked data from students at UC Berkeley, UC Davis, and K-12 schools, affecting approximately 18,000 users. The incident highlights the importance of security in AI application development and the potential risks involved, urging developers to prioritize security measures when building AI tools.