Netskope

Clicks on phishing links decreased by 27%, from 119 per 10,000 users last year to 87 per 10,000 users this year.

60% of insider threat incidents involved personal cloud application instances in 2025.

The average organization saw a twofold increase in data policy violations related to generative AI applications over the past year.

The average organization experienced 223 incidents of data policy violations related to generative AI applications each month from October 2024 to October 2025.

The percentage of AI users utilizing personal AI applications decreased from 78% to 47% from 2024 to 2025.

87 out of every 10,000 users clicked on a phishing link each month in 2025.

The percentage of users uploading data to personal cloud applications increased by 21% over the past year, with 31% of users in the average organization uploading data monthly.

12% of organizations detected employee exposure to malware via GitHub each month in 2025.

Regulated data accounted for 54% of data policy violations linked to personal cloud applications in 2025.

The top 25% of organizations experienced an average of 2,100 data policy violation incidents per month across 13% of their generative AI user base in 2025.

The number of organizations implementing real-time controls on data sent to personal applications increased from 70% to 77% from the previous year.

The number of users utilizing SaaS generative AI applications tripled in the average organization from October 2024 to October 2025.

LM Studio is used by 0.9% of organisations.

In the three months ended May 2025, users of genAI platforms increased by 50%.

There has been a 50% spike in genAI platform usage among enterprise end-users in the three months ended May 2025.

34% of organisations are currently using Large Language Model (LLM) interfaces.

In May 2025, 41% of organisations were already using at least one genAI platform.

Ollama is used by 33% of organisations.

GitHub Copilot is now used in 39% of organisations.

Over half of all current app adoption among enterprise users is estimated to be shadow AI.

Netskope is tracking more than 1,550 distinct genAI SaaS applications, which is up from just 317 in February.

Network traffic tied to genAI platform usage increased 73% over the prior three-month period.

Approximately 29% of organisations are utilising Microsoft Azure OpenAI.

Ramalama is used by 0.6% of organisations.

5.5% of organisations have users running agents generated from popular AI agent frameworks on-premises.

13% of organisations have users making API calls to api.anthropic.com.

Two-thirds (66%) of organisations have users making API calls to api.openai.com.

Organisations are now using approximately 15 genAI apps, up from 13 in February 2025.

The amount of data uploaded to genAI apps each month has increased from 7.7 GB to 8.2 GB quarter over quarter.

ChatGPT saw its first decrease in enterprise popularity since Netskope started tracking it in 2023.

Grok has entered the top 10 most-used applications for the first time.

Users are downloading resources from Hugging Face at a majority (67%) of organisations.

22% of organisations are utilising Amazon Bedrock.

Grok remains in the top 10 most-blocked apps list

7.2% of organisations are utilising Google Vertex AI.

Industries with the lowest AI adoption rates that will see increases next year include banking, state governments, local governments, and education. Even in these industries, adoption is already high, at or above 85%.

DLP adoption varies widely by industry, with the telecommunications sector leading all others at 64% DLP adoption for genAI.

At the current trajectory, 96% of organizations will be using genAI apps by the end of 2025.

The top 25% of organizations had at least 21% of their people using genAI apps, while the bottom 1% had just 1.7%.

Other types of data involved in policy violations include: intellectual property (16%), source code (13%), passwords and keys (11%), and encrypted data (1%).

The list of the top 10 most blocked AI apps are: QuillBot (33%), Beautiful.ai (31%), AiCHatting (30%), Pixlr (28%), Tactiq (27%), Writesonic (27%), DeepAI (24%), ElevenLabs (24%), Craiyon (24%), and Poe AI (23%)

Real-time, interactive user coaching is used in 34% of organizations to control genAI data risk by empowering individuals to make informed decisions about AI risk in real time.

The top apps for upstream activities to personal apps are Google Drive (95%), OneDrive (92%), LinkedIn (91%), Facebook (90%), Google Calendar (89%), Gmail (88%), ChatGPT (86%), Twitter/X (86%), Outlook.com (82%), and Google Gemini (78%).

In 2024, downloads of malicious content from popular cloud apps occurred in 88% of organisations at least once per month.

73% of organisations block at least one GenAI app, with a steady rate of 2.4 GenAI apps blocked on average year over year.

The top phishing targets by links clicked are cloud services (27%), banking (17%), telecommunications (13%), social media (11%), and government (10%).

The top 25% of retail organizations had at least 34% of people using genAI apps, while the top 25% in technology led all other industries with at least 41% of people using genAI apps. At the other end of the spectrum, banking trailed with only 3% of users in the average organization using genAI.

While 94% of organizations are using genAI apps, more than 99% of organizations have controls in place to mitigate the risks that genAI apps pose.

1.4 out of every 100 people encountered malicious content on the web or in the cloud each month during 2024. The most common types of content encountered were JavaScript-based Trojans.

The percentage of data policy violations in organizations is as follows: Source Code (40%), Regulated Data (32%), Intellectual Property (15%), and Passwords and Keys (13%)