人工智能在全球范围内越来越普及,但管理不善会危及数据安全. AI adoption rises globally, but weak governance risks data security.

在南非和全球,67%的公司采用人工智能,比2024年的45%有所增加,但只有14%的公司有正式的人工智能战略,32%的公司非正式地使用工具,造成数据泄漏和模型操纵等安全风险。 Generative AI adoption is surging in South Africa and globally, with 67% of companies using it—up from 45% in 2024—yet only 14% have formal AI strategies, and 32% use tools unofficially, creating security risks like data leakage and model manipulation. 网络安全专家警告说,能够迅速扩大规模和以非决定性方式运作的AI代理商缺乏适当的身份管理和监督,只有10%的组织拥有治理框架。 Cybersecurity experts warn that AI agents, which can scale rapidly and operate non-deterministically, lack proper identity management and oversight, with only 10% of organizations having governance frameworks. 这一转变正在将网络风险从基础设施转移到数据层,对传统安全模式提出了挑战。 This shift is moving cyber risks from infrastructure to the data layer, challenging traditional security models. 像Check Point和Okta这样的公司正在推广安全的AI实践,包括身份控制,准时访问和与欧盟AI法等法规相一致的治理框架. Firms like Check Point and Okta are promoting secure AI practices, including identity controls, just-in-time access, and governance frameworks aligned with regulations like the EU AI Act. 由于AI驱动的工作流程绕过标准防御系统,要求加强数据治理、工作人员培训和咨询作用,以管理不断变化的威胁,因此缔约国面临越来越多的问责制。 MSPs face growing accountability as AI-driven workflows bypass standard defenses, requiring stronger data governance, staff training, and advisory roles to manage evolving threats.