Research reveals growing gaps between AI experimentation and risk patterns as usage deepens across business workflows<\/p>\n
\u00a0Enterprise use of AI is expanding rapidly across development, operations, and knowledge work, but new research shows these changes in behavior are creating new data risks that legacy technology can\u2019t see or govern. The\u00a02026 AI Adoption & Risk Report, released today by Cyberhaven Labs and based on billions of real-world data movements around Generative AI SaaS applications, endpoint AI applications, and AI agents, offers a clear-eyed look at how enterprises are actually using AI and why governance and data security must be ready for these changes.<\/p>\n
\u201cWhat this research makes clear is that enterprise AI adoption isn\u2019t just accelerating, it\u2019s fragmenting,\u201d said Nishant Doshi, CEO of Cyberhaven. \u201cA small set of teams is moving fast and embedding AI deeply into daily work, while security and governance are often playing catch-up. As organizations plan for 2026 and beyond, the risk isn\u2019t AI itself; it\u2019s not understanding how AI is actually being used. Without visibility into which tools are in play, what data is flowing through them, and where controls need to adapt, enterprises risk widening the gap between innovation and trust.\u201d<\/p>\n
The report surfaces the following key findings about enterprise AI adoption and usage:<\/p>\n
1. An AI Adoption Gap is EmergingAI adoption and use is not unfolding as a steady, industry-wide wave. Instead, it is becoming increasingly polarized.<\/p>\n
A widening gap is emerging between AI early adopters and organizations that remain hesitant to embrace these technologies.<\/p>\n
The top 1% of early adopter organizations use more than 300 GenAI tools.<\/p>\n
By contrast, cautious enterprises typically employ fewer than 15 GenAI tools.<\/p>\n
2. Most GenAI SaaS Tools Are Objectively Risky, with the Average Employee Entering Sensitive Data into AI Tools Once Every Three DaysMost AI use today occurs in tools that do not meet traditional enterprise risk standards \u2014 yet employees continue to enter sensitive data into them at high rates.<\/p>\n
Across the top 100 most-used GenAI SaaS applications, 82% are classified as \u201cmedium,\u201d \u201chigh,\u201d or \u201ccritical\u201d risk.<\/p>\n
Cyberhaven Labs data shows that 32.3% of ChatGPT usage occurs through personal accounts, as does 24.9% of Gemini usage.<\/p>\n
39.7% of all data movements into AI tools involve sensitive data, including prompts or copy-paste actions.<\/p>\n
This behavior significantly limits organizational visibility into AI usage and data flows.<\/p>\n
3. Coding Assistants and AI Agents are Becoming the \u201cSecond Wave\u201d of Workplace AIAI coding assistants (such as Cursor, GitHub Copilot, and Claude Code) continued to grow steadily through 2025.<\/p>\n
In companies leading in AI adoption, nearly 90% of developers use these tools, whereas in a typical organization, adoption is closer to 50%.<\/p>\n
On the other side of the spectrum, only 6% of developers use AI coding assistants. This illustrates the growing gap of AI adoption, with developers at frontier companies being 11.5\u00d7 more likely to use AI coding assistants.<\/p>\n
In the later months of 2025, 30% of developers using AI coding assistants reported using at least two.<\/p>\n
As enterprise AI adoption continues to accelerate, the\u00a0Cyberhaven Labs 2026 AI Adoption & Risk Report\u00a0underscores a widening divide between innovation and oversight. AI adoption is becoming uneven across organizations, teams, and workflows, with the highest levels often occurring in environments with the least mature governance and visibility.<\/p>\n
\u201cAI is no longer a side experiment for most enterprises; it\u2019s becoming a core part of the infrastructure,\u201d added Doshi. \u201cOrganizations that succeed will be those that move beyond one-size-fits-all policies and invest in security approaches that reflect real usage patterns. By bringing visibility, context, and control together, enterprises can enable teams to innovate with AI while maintaining trust, compliance, and resilience as adoption continues to evolve.\u201d<\/p>\n
Explore the full report findings\u00a0here.<\/p>\n
These research findings will be discussed live during a webinar today with Harvard Business Review Analytic Services, featuring a conversation among Alex Clemente, Managing Director, HBR-AS, Nishant Doshi, CEO, Cyberhaven, and Dan Walsh, CISO, Datavant. Register\u00a0here.<\/p>\n
Earlier this week, Cyberhaven announced the\u00a0general availability of its Data Security Posture Management solution, a key part of its unified AI & data security platform, purpose-built to protect sensitive data wherever it lives and goes, including endpoints, SaaS, cloud, on-prem environments, and AI workflows.The post Cyberhaven Research Highlights Urgent Need for AI Data Governance first appeared on AI-Tech Park.<\/p>\n","protected":false},"excerpt":{"rendered":"
Research reveals growing gaps between AI experimentation and risk patterns as usage deepens across business workflows \u00a0Enterprise use of AI is expanding rapidly across development, operations, and knowledge work, but new research shows these changes in behavior are creating new data risks that legacy technology can\u2019t see or govern. The\u00a02026…<\/p>\n
The post Cyberhaven Research Highlights Urgent Need for AI Data Governance<\/a> first appeared on AI-Tech Park<\/a>.<\/p>\n<\/div>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"site-container-style":"default","site-container-layout":"default","site-sidebar-layout":"default","disable-article-header":"default","disable-site-header":"default","disable-site-footer":"default","disable-content-area-spacing":"default","footnotes":""},"categories":[4,1],"tags":[3],"class_list":["post-1101","post","type-post","status-publish","format-standard","hentry","category-ai","category-ai-and-ml","tag-ai"],"yoast_head":"\n