印度启动了5个AI安全项目,以打击深刻的假象、偏见和安全威胁。 India launches five AI safety projects to combat deepfakes, bias, and security threats.

印度在其印度AI特派团下启动了五个新的AI安全项目,重点是深层假发现、减轻偏见和AI安全。 India has launched five new AI safety projects under its IndiaAI Mission, focusing on deepfake detection, bias mitigation, and AI security. 从400多项提案中挑选出来的倡议包括一个多试剂检测错误信息的系统、可解释的伪造内容的人工智能、实时声音深层工具、农业人工智能中的性别偏见评估,以及大型语言模型的渗透测试平台。 Selected from over 400 proposals, initiatives include a multi-agent system for misinformation detection, explainable AI for forged content, real-time voice deepfake tools, gender bias assessments in agricultural AI, and a penetration testing platform for large language models. 这些努力旨在加强使用符合道德的人工智能和国家技术抗御能力。 The efforts aim to strengthen ethical AI use and national technological resilience.