The adoption of artificial intelligence tools is soaring across the retail sector, but this rapid technological expansion has brought significant visibility challenges and heightened risks to regulated consumer data, according to a comprehensive new report from security firm Netskope.
While retail organizations have made strides in reining in rogue technology use and steering staff toward officially sanctioned platforms, the proliferation of hidden AI integrations, autonomous software agents, and sophisticated cybercriminal lures continue to threaten enterprise security. The findings, drawn from data collected between July 1, 2025, and July 30, 2026, highlight the complex balancing act retailers face as they attempt to harness automation while protecting sensitive information.
Gaining Control Over Standalone AI Tools
For years, the retail industry struggled with shadow AI—the unauthorized use of generative applications and large language models by employees working without IT oversight. However, recent data indicates that retailers have successfully tightened their grips on these workflows.
According to Netskope’s findings, the share of retail employees using ungoverned AI tools plummeted significantly, dropping from 70% down to 44% compared to the firm’s previous annual report. Concurrently, the proportion of employees utilizing officially approved corporate tools surged from 40% to 73%. Overall, the research shows that roughly two-thirds of all retail employees now interact with some form of standalone AI tool as part of their daily routines.
This shift reflects a concerted effort by corporate IT and security teams to implement centralized management structures rather than leaving software acquisition up to individual departments or workers. By providing secure, sanctioned alternatives, retailers have successfully drawn workers away from unsecured consumer-grade applications.
The Hidden Threat of Integrated AI and Customer Data
Despite the positive trend toward centralized management, security experts warn that standalone AI applications represent only one piece of a much larger puzzle. Nearly all retail employees surveyed reported using broader enterprise software that contains built-in AI features, and 90% stated they routinely use tools trained directly on customer data.
As this adoption continues to grow, researchers noted that retailers face a much greater challenge in understanding precisely where sensitive information is being shared and how it might be utilized. This exposure happens not only through direct employee interactions with chatbot interfaces, but also through automated AI functionalities operating quietly in the background of standard enterprise platforms.
These visibility gaps can carry severe consequences for retail organizations if a lack of oversight results in the exposure of confidential customer information. Among all the AI-related data policy violations tracked by Netskope, an overwhelming 56% involved data protected by federal or state regulations. Furthermore, additional violations included the exposure of proprietary source code, which accounted for 20% of incidents, and the leakage of sensitive passwords or application programming interface (API) keys, which represented 16% of tracked infractions.
In competitive consumer markets, a major customer data breach can trigger immediate and severe reputational damage alongside heavy regulatory penalties, making comprehensive visibility an urgent operational priority.
The Rapid Rise of Agentic AI and Remote Connections
Beyond traditional conversational models, agentic AI—autonomous software systems capable of executing complex, multi-step workflows on behalf of users—has emerged as a major security concern within the retail sector. Many businesses currently lack the technological capability to track precisely which remote servers these autonomous agents are accessing as they perform their background tasks.
Netskope reported observing a staggering 400% increase in the number of retail-sector AI agents interacting with remote Model Context Protocol (MCP) servers during the survey period. This dramatic growth underscores a critical vulnerability, as remote MCP connections introduce complex new pathways for data to move fluidly between internal AI applications and external systems.
For modern retailers, establishing strict visibility and absolute control over these machine-to-machine interactions has become paramount, especially in operational scenarios where autonomous AI agents are granted direct access to core business data and other sensitive enterprise resources.
Phishing Lures and Ongoing Defenses
As employees continue actively searching for and experimenting with new productivity-enhancing AI tools, malicious actors have adapted their tactics by designing sophisticated phishing lures that expertly mimic trusted AI applications.
Data from the report indicates that this malicious phishing activity experienced a temporary decline between May 2025 and December 2025 before picking up momentum again in early 2026. By March, AI-related phishing lures reached a density of roughly 100 targeted users per 100,000. Security researchers emphasize that this targeted phishing strategy remains an ongoing, persistent threat as long as employees continue their widespread search for emerging AI applications.
To combat these overlapping security crises, Netskope recommends that retail organizations implement a multi-layered defense strategy. Recommended measures include proactively blocking employee access to unnecessary or unverified applications, continuously inspecting encrypted web traffic, and deploying robust data-loss-prevention policies specifically designed to detect sensitive corporate and customer information before it can be transmitted to ungoverned AI environments.
