Security Fears Stall AI Expansion as Attack Surface Grows

Security complexity emerges as primary AI adoption barrier

Enterprise technology leaders are hitting the brakes on AI expansion, citing mounting cybersecurity risks as networks strain under the weight of new AI workloads. Security complexity has emerged as the single most significant challenge tied to AI-driven network demand, according to a recent Cisco report based on a global survey of 3,472 CIOs and technology executives.

The findings reveal a direct tension: organizations recognize AI's potential but fear that scaling deployments will expose them to unacceptable hacking risks. This security hesitation is creating what Cisco characterizes as "a direct barrier to AI scale."

Why it matters

The security-adoption paradox threatens to slow enterprise AI momentum at a critical juncture. If organizations can't confidently secure AI workloads, the technology's business value remains theoretical. Meanwhile, 71% of IT leaders believe AI-driven threats will evolve faster than their defenses can adapt—a gap that could widen as adversaries weaponize the same tools enterprises are struggling to protect.

Network modernization driven by security concerns

Security risks are pushing infrastructure upgrades across enterprises. Seventy-two percent of IT leaders cited increased security risks or an expanded attack surface as a key reason for modernizing their networks to support AI tools, according to the Cisco research conducted between March and April 2025.

More than three-quarters of respondents expect security risks to intensify as AI use cases move beyond generative applications into operational systems. A similar proportion reported that AI has already expanded their attack surface within the past year.

"We're just playing catch-up at the moment," a senior IT executive in the U.K. education sector told Cisco researchers.

Visibility gaps and shadow AI complicate defenses

Nearly 70% of survey respondents reported growing blind spots across their networks, hampering their ability to detect and block suspicious activity. The proliferation of AI tools has created what Cisco describes as "shadow AI activity"—unauthorized or unmonitored AI deployments that operate outside security team oversight.

One retail executive described the challenge: "The issue from a security standpoint is that it's hard to create the guardrails for every possible AI tool that your organization must use."

The report identified expanded attack surfaces, inconsistent policy enforcement, and limited visibility into AI-driven network traffic as key factors driving organizational hesitation around AI adoption.

Security controls lag behind deployment pace

While nearly 90% of IT leaders have implemented security controls for AI tools, confidence remains shaky. Sixty-one percent said they're waiting for greater confidence in their security posture before expanding AI use across their organizations.

The challenge is compounded by AI's operational speed. Systems that generate continuous activity across distributed networks can transform minor security gaps into significant governance exposures, Cisco noted.

The findings were first reported by Cybersecurity Dive, drawing on Cisco's survey of technology leaders worldwide.