Generative AI tools operate within almost every workplace in Europe, embedded in meeting transcription services, writing assistants, coding co-pilots, and search functions. Employees in the region incorporate these tools, including customer records, financial information, and proprietary code, into their daily work, and their high volume of activity has created a measurable pattern in where data breaches occur. of Netskope Threat Labs Report: Europe in 2026 has documented this pattern across European organizations over the past year.
Source: Netskope Threat Labs
Almost complete adoption with changing governance
The use of AI reaches around 99% of organizations in Europe, and the proportion of individual users actively interacting with AI applications has risen from 35% to 65% in the past year. Direct interactions with chatbots and assistants only tell part of the story. Currently, approximately 95% of users indirectly use applications that incorporate AI-powered functionality, and 89% interact with applications that rely on user data for training.
European companies are moving to a licensed environment. Over the same period, the percentage of users with personal AI accounts fell from 79% to 43%, and the percentage of users with organization-managed AI solutions increased from 28% to 72%. The opposite trend complicates the situation. The percentage of users switching between personal and corporate accounts has increased from 7% to 15%, indicating continued shadow AI activity even as governance programs mature.
Regulated data accounts for the majority of exposure incidents
Data policy violations across AI and personal cloud applications are concentrated in regulated information, accounting for 59% of incidents. Source code comes in at 15%, intellectual property at 13%, and passwords and API keys at 12%. This pattern shows compliance-sensitive materials as the category most often pushed to AI tools or personal cloud accounts in a way that triggers data loss prevention rules.
ChatGPT leads, Claude surpasses Gemini
The European application composition differs from the world ranking. ChatGPT remains the most widely used AI service across the region, with Anthropic Claude holding second place ahead of Google Gemini. This order reverses the global pattern in which Gemini is placed before Claude. French-developed assistant Mistral Le Chat also joins the regional mix.
Claude’s rise accelerated in September 2025, and his adoption curve suddenly steepened, overtaking Gemini. ChatGPT remained in the top spot throughout the year, and Microsoft Copilot maintained steady usage.
Blocked applications reflect privacy concerns
Many organizations restrict certain AI applications that they consider risky. Particular Audience topped the block list with 44%, followed by ZeroGPT with 37% and DeepSeek with 36%. Application drawing blocks raise questions about transparency of data processing, how personalization works, and visibility into how user data is processed and retained. In regulated areas, comprehensive category blocks supplement individual app controls.
Attackers slip into trusted cloud services
Malware distribution in Europe relies on widely trusted cloud platforms. Attackers continue to host malicious payloads on services such as GitHub and Microsoft OneDrive, which provide reputational trust and often bypass URL-based filtering. Using personal cloud applications within a corporate network blurs the lines between work and personal data flows, opening up new avenues that users are exposed to when moving files between environments.
What the data shows
In a year, the European organization built guardrails around AI and moved most users from personal accounts to sanctioned platforms. The remaining work focuses on three pressure points. For one, 15% of users still switch between personal and business accounts, AI capabilities built into everyday productivity tools that operate below the visibility threshold of many security programs, and a steady traffic of malicious files arriving via trusted cloud storage.
Netskope Threat Labs recommends combining data loss prevention controls with application-specific governance. This is because regulated data breaches occur in both AI services and personal cloud applications, and the line between the two categories becomes thinner as AI capabilities are included in all major productivity suites.
Download: IT and Security Guide to AI Deployment
