当前位置:首页 >英文主页 >中英对照 > 中译版报告详情

Cyberhaven :2026年企业AI应用与风险报告(中译版)(20页).pdf

上传人: 1****1 编号:1114192 2026-02-09 20页 440.83KB

下载:

1、2026 AI Adoption&Risk ReportCyberhaven Labs22026 AI Adoption&Risk ReportTable ofContentsIntroduction 3Key Findings4Section 1:An AI Adoption Gap is Emerging5Section 2:Enterprise AI Usage Spans Models,Tools,and Accounts Outpacing Governance7Section 3:Regulated Industries Are Adopting AI Most Frequentl

2、y12Section 4:Employees Are Pouring Sensitive Data into Risky AI tools14Section 5:Coding Assistants and AI Agents are Becoming The“Second Wave”of Workplace AI16Conclusion1932026 AI Adoption&Risk ReportIntroductionSince the launch of ChatGPT in 2022,AI has become one of the fastest-adopted workplace t

3、echnologies in history.What began as individual employees experimenting with generative AI has rapidly evolved into tools embedded directly into core business workflows across organizations of all sizes.The speed of this transition,from novelty to operational dependency,has been unprecedented.In man

4、y cases,it has outpaced the ability of enterprises to understand,govern,and secure AI usage.AI adoption has not progressed evenly across the enterprise.Usage is increasingly polarized.A small but growing set of organizations is moving aggressively,deploying dozens or even hundreds of AI tools across

5、 development,operations,and knowledge work.Others remain cautious or only lightly engaged.In high-adoption environments,innovation often advances faster than governance.Experimentation frequently takes priority over visibility,control,and risk management.This years research from Cyberhaven Labs capt

6、ures a clear shift in how enterprises are using AI and where risk is concentrating as a result.To understand the full scope of adoption,we analyzed AI usage across three categories:Generative AI SaaS applications,endpoint AI applications,and AI agents.Drawing on billions of real-world data movements

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
1. **AI采用差距扩大**:领先企业采用超300个AI工具,员工采用率71.4%,而谨慎企业不足15个工具,采用率低至2.5%。 2. **中国开源模型流行**:端侧中国开源模型占比50%,如DeepSeek、Qwen,带来数据主权与合规风险。 3. **工具风险高**:82%的常用GenAI SaaS工具被列为“中高”或“关键”风险,39.7%的交互涉及敏感数据(如代码、财务信息)。 4. **个人账户使用普遍**:32.3%的ChatGPT、60.9%的Perplexity交互通过个人账户,增加数据泄露风险。 5. **第二波AI浪潮**:49.5%开发者使用AI编码助手,领先企业达90%;23%企业采用定制AI代理,制药、制造行业领先。
AI风险有多高? 谁在用AI工具? 数据安全怎么办?
客服
商务合作
小程序
服务号
折叠