Enterprise AI and ML is on the rise, but security is now being put to the test to see if it can keep up with the emerging tech’s rapid advancement.

A new report from Zscaler makes clear that the power and perils of AI loom larger than ever, as threat actors continue to push the boundaries of AI’s malicious capabilities.

The Zscaler report analysed 536.5 billion transactions from AI/ML tools between February and December 2024, revealing surprising but on trend shifts in enterprise usage worldwide.

The report saw that AI and machine learning tool usage saw an exponential year-on-year rise, with 36 times more transactions from over 800 AI?ML applications in the Zscaler cloud. This highlights the explosive growth in enterprise interest – as well as growing dependence – on these technologies.

ChatGPT remains the top application used by transaction volume with nearly half (45.2%) of all transactions, despite ongoing debates over its security implications.

Enterprises blocked 59.9% of all AI/ML transactions, reflecting concerns around AI data security and the steps companies are taking in shaping their approaches to AI governance.

ChatGPT was also the most-blocked AI application among known applications, followed by Grammarly, Microsoft Copilot, QuillBot, and Wordtune, reinforcing growing interest and caution when it comes to AI-powered writing and productivity assistants in enterprise settings.

The growing risk of AI adoption is highlighted by the sheer amount of data being transferred via these AI tools which amounted to 3624 TB according to the report.

tiatives.

Recommended reading

• Cloud Spending Surged By 21% Last Quarter
• Why Cloud Is the Backbone of Enterprise AI Adoption
• Caution Over Cloud Adoption Is Holding UK Businesses Back

Globally, 28% o

Why Cloud Is the Backbone of Enterprise AI Adoption

When it comes to sectors, finance & insurance, as well as manufacturing, generate the most AI/ML traffic, with 28.5% and 21.6% share of all AI/ML transactions in the Zscaler cloud. Services (18.5%), technology (10.1%), healthcare (9.6%), and government (4.2%) trailed significantly.

The report also highlighted that AI continually amplifies cyber risks, fueled by advancements in deepfake technology, emerging open source AI models, and autonomous attack automation – undoubtedly making threats more adaptive, targeted, and difficult to detect.

The report highlights the double-edged sword of AI proliferation across enterprises.