2026年人工智能应用与风险报告

2026 AIAdoption &Risk Report Cyberhaven Labs Table of Introduction Key Findings Section 1:An AI Adoption Gap is Emerging Section 2:Enterprise AI Usage Spans Models, Tools, and Section 3:Regulated Industries Are Adopting AI Most Frequently Section 4:Employees Are Pouring Sensitive Data into Risky AI tools Section 5:Coding Assistants and AI Agents are Becoming Conclusion Introduction Since the launch of ChatGPT in 2022, AI has become one ofthe fastest-adopted workplace technologies in history. plateau, enterprise AI adoption is not slowing down.In 2025, AI coding assistants, browser-basedagents, and custom AI agents saw rapid growth.This second wave of adoption is more operational What began as individual employees experimentingwith generative AI has rapidly evolved into toolsembedded directly into core business workflowsacross organizations of all sizes. The speed of this AI adoption has not progressed evenly acrossthe enterprise. Usage is increasingly polarized. Asmall but growing set of organizations is movingaggressively, deploying dozens or even hundredsof AI tools across development, operations, and As AI becomes infrastructure rather than astandalone interface, the security implicationsintensify. Employees are no longer using AI only forideation or research. They are inputting source code,financial data, customer information, and intellectual This year’s research fromCyberhaven Labs captures aclear shift in how enterprisesare using AI and where risk is The result is a familiar pattern, amplified by scaleand speed. Shadow adoption increases. Controlsare applied inconsistently. Risk accumulates fasterthan most organizations can measure or manage it. To understand the full scope of adoption, weanalyzed AI usage across three categories:Generative AI SaaS applications, endpoint AIapplications, and AI agents. Drawing on billionsof real-world data movements from hundreds This report provides a data-driven view into howenterprises used AI in 2025, where adoptionis accelerating, and where security risk iscompounding. By examining real-world usage The data shows that while usage of traditionalchat-based GenAI SaaS tools is beginning to Key Findings Organizations with the highest rates of AIadoption are utilizingover 300GenAI tools 02 based usage among Cyberhaven users. 03 are classified as “medium,” “high,” or “critical” risk.04 One-thirdof employees are accessingGenAI tools from personal accounts,increasing overall risk and Shadow AI. Employees are feeding AI tools sensitivedata, as over a third (39.7%) of all interactionswith AI tools involve sensitive data. 1 An AI AdoptionGap is Emerging Artificial intelligence (AI) andlarge language models (LLMs) are Today, 62% of organizations1are experimenting with AI agents,enterprises are spending four times more on AI software thanon traditional software, and 74% of executives stated2they However, AI adoption and useis not unfolding as a steady,industry-wide wave. Instead, it is A widening gap is emerging between AIearly adopters and organizations that remain Frontier enterprises — those with the highest rates of AIadoption — are interacting with hundreds of GenAI applicationsover the course of 2025. In the most advanced cases,organizations are using more than 300 GenAI tools, while even This divide is stark when compared to the medianorganization, which uses 54 GenAI applications. Inpractice, frontier enterprises are adopting AI tools at GenerativeAI SaaS The same divide appears at the employee level. In the average organization, roughly one-third of employees use GenAI tools regularly. Yet adoption rates varydramatically by enterprise maturity. Frontier organizations see a 71.4% employee adoption rate, while the mostcautious enterprises report adoption as low as 2.5%. As organizations deploy more GenAI tools, employee usage As discussed in Section 3, industry-level variance likely contributes to this wide distribution, with certain sectorsaccounting for a disproportionate share of GenAI usage while others lag behind. Most Organizations RemainHesitant to Adopt AI As data flows through hundreds of GenAI tools,rapid adoption multiplies risk points, governancecomplexity, and potential sensitive data exposure. Manyorganizations appear to be trading coordination andsecurity controls for experimentation, creating a growinggap between AI adoption and AI security. This challenge While frontier organizations are rapidly experimentingwith GenAI, the majority of enterprises remain cautious.Within the median organization, only 33.4% of employees This polarized adoption pattern reveals two realities.Some organizations are aggressively adopting AI andmay realize outsized gains in innovation and growth.At the same time, these frontier enterprises are also 2 Enterprise AI Usage SpansModels, Tools, and Accounts Chinese Open-WeightModels Growing in While AI is often synonymous with “ChatGPT” inpopular discourse, much as “Google” becamesh