AI Adoption Soars 91%, Security Risks Intensify
▼ Summary
– AI adoption is outpacing enterprise oversight, creating a rapidly expanding attack surface across all sectors.
– The report analyzed nearly 990 billion AI/ML transactions from 2025, showing a 91% growth in AI usage across over 3,400 applications.
– Many organizations lack a basic inventory of their AI models and embedded AI features, despite this massive growth.
– Finance and insurance are the most AI-driven sectors by volume, while technology and education saw the fastest growth in AI transactions.
– Engineering departments accounted for nearly half of all AI usage, followed by IT and marketing.
The rapid integration of artificial intelligence into business operations is outpacing the ability of many organizations to manage it securely, exposing them to a broader and more complex set of cyber threats. A new analysis of nearly a trillion transactions reveals a staggering 91% surge in AI application usage, yet a concerning number of companies operate without a fundamental inventory of the AI models and features they employ. This gap between adoption and governance is creating significant vulnerabilities that malicious actors are increasingly eager to exploit.
Finance and insurance continue to lead in AI implementation by sheer volume, generating nearly a quarter of all observed AI and machine learning traffic. However, the most dramatic growth rates belong to the technology and education sectors, which saw their AI transactions skyrocket by 202% and 184% respectively over the past year. This explosive expansion underscores the technology’s pervasive reach across different industries.
Within organizations, the distribution of AI usage is highly concentrated. Engineering departments are the primary drivers, accounting for almost half of all AI activity. Information technology teams follow as the second-largest users, responsible for nearly a third of usage, while marketing functions represent a smaller but still significant portion. This departmental focus highlights how AI is being leveraged for core product development, infrastructure management, and customer engagement.
The security implications of this unchecked growth are severe. The lack of a centralized inventory means security teams often cannot see or secure the AI tools their colleagues are using. Shadow AI, the unauthorized use of applications, becomes a major concern, as employees may utilize generative AI tools that could inadvertently leak sensitive corporate data or intellectual property. Furthermore, each new AI model or application represents a potential entry point for attackers, whether through compromised APIs, data poisoning attacks, or maliciously crafted prompts designed to manipulate the AI’s output.
To mitigate these risks, experts recommend a shift toward consolidated and secure AI access. Implementing a Zero Trust architecture that rigorously inspects all AI traffic is becoming essential. This approach ensures that every request is authenticated and authorized, regardless of its origin, preventing data exfiltration and blocking threats hidden within AI transactions. Companies must also establish clear acceptable use policies, provide secure, vetted alternatives to risky public AI tools, and invest in employee education to foster a culture of secure innovation.
(Source: InfoSecurity Magazine)
