Palo Alto Networks: 890% jump in GenAI use driving new wave of cyber incidents
Organizations now use dozens of AI tools, many with little or no IT oversight, the report finds.
The use of generative artificial intelligence (GenAI) applications in the business sector surged by 890% last year, even as the number of data security incidents linked to these tools rose 2.5 times, according to Palo Alto Networks' annual State of Generative AI report, published this week.
The report is based on traffic data from Palo Alto Networks' servers, tracking GenAI application usage among approximately 7,000 customers. Data security incident data was collected between January and March 2025. The most striking figure is the 890% year-over-year increase in GenAI usage by organizations, most of which occurred in the final quarter of 2024.
On average, 66 different GenAI applications are used per organization, with some organizations having hundreds of applications. “Adoption and usage spikes around the launch or update of AI models, as businesses and users evaluate the impact of the latest developments on their productivity and innovation needs. For example, given the increased demand following the launch of DeepSeek-R1 in January 2025, we saw a significant 1,800% spike in DeepSeek traffic in the two months that followed.”
“The notable surge in October may suggest organizations bringing more GenAI initiatives to production,” the report notes. “It’s no surprise that GenAI apps are being leveraged at a record pace. As with many consumer applications, they are also being introduced into enterprise workspaces - with or without the approval or oversight of IT departments.”
On average, organizations use 66 different GenAI applications, with some deploying hundreds. Adoption and usage spike around the launch or update of AI models, as businesses assess how new capabilities can enhance productivity and innovation, the report explains.
The most commonly used GenAI tools are writing assistance applications (34% of total usage), which help draft emails, posts, and reports. Chatbots come next with 28.9%, followed by internal organizational search tools (10.6%) and developer platforms (10.3%), which assist with code generation. Together, these four categories account for 83.8% of all GenAI usage in enterprises.
However, the rapid rise in adoption brings significant security risks. According to Palo Alto Networks, organizations use an average of 6.6 GenAI applications deemed “high-risk.” Correspondingly, data leakage incidents tied to GenAI apps, detected and prevented by the company, rose 2.5-fold in Q1 2025, and now account for 14% of all identified incidents.
