微软识别了100多种AI产品的安全缺陷,Cisco预测AI将占主导地位。 Microsoft identifies security flaws in over 100 AI products, as Cisco predicts agentic AI will dominate.

微软的人工智能红队发布了一份白皮书, 详述100多种人工智能产品的脆弱性, Microsoft's AI red team has released a whitepaper detailing vulnerabilities in over 100 generative AI products, highlighting new risks like prompt injection and traditional issues such as outdated software. 与此同时,Cisco预测人工智能将占主导地位,提供先进的解决问题能力,但引起对安全和数据处理的担忧。 Meanwhile, Cisco predicts that agentic AI will dominate, offering advanced problem-solving capabilities but raising concerns about security and data handling. 专家们强调,需要不断进行测试和人力监督,以便有效地确保AI系统的安全。 Experts stress the need for continuous testing and human oversight to secure AI systems effectively.