UK Employees’ Reckless Use of Unmanaged AI Creates Significant Security Blind Spot, Report Warns

rifanmuazin - Thursday, November 19, 2026 | 13:10
Finance Poetry
UK Employees’ Reckless Use of Unmanaged AI Creates Significant Security Blind Spot, Report Warns

A growing wave of UK employees is inadvertently exposing their organizations to severe security risks by sharing sensitive information with unmanaged AI tools. A new report from credential security specialist Dashlane, titled Misplaced Trust: Inside the Workplace’s New AI Security Blind Spot, reveals that workers are increasingly granting AI tools and autonomous agents access to corporate systems and sensitive data without considering the potential consequences for their employers.

The findings are based on an extensive survey of 3,800 workers and IT executives across the UK, US, and France, including a focused subset of 1,000 UK employees and 250 UK IT executives. The data highlights a troubling disconnect between the rapid, often clandestine adoption of generative AI in the workplace and the security protocols necessary to govern such powerful, and potentially volatile, technology.

The Growing Prevalence of Shadow AI

Dashlane’s research paints a concerning picture of how modern employees interact with AI. According to the report, nearly one in five UK workers (19%) have shared highly sensitive credentials—such as passwords, API keys, or specific login details—with an AI tool or agent. Even more alarming is the discovery that nearly half of the respondents (48%) have permitted AI tools to execute tasks without prior authorization. These autonomous actions frequently include accessing internal emails, sensitive company files, and private customer information, effectively granting these third-party tools a "key to the kingdom" without the explicit oversight of IT departments.

This trend is a clear manifestation of "shadow AI," a phenomenon where employees utilize AI-powered software, browser extensions, or platforms that have not been vetted or approved by their organization’s IT or security teams. When employees engage in shadow AI, they often do so out of a desire for increased productivity or convenience, unaware that their actions could be bypassing essential security controls.

Christophe Frenet, Chief Product Officer at Dashlane, describes this trend as the "AI Colleague Effect." This phenomenon occurs when employees extend an unwarranted level of trust to AI systems, treating them as reliable, sentient coworkers rather than as sophisticated software models that operate on external servers. "Employees are overextending trust to AI systems, sharing credentials and granting access to AI agents, sometimes without fully understanding what those tools can do with the permissions they’ve been given," Frenet explained. "That creates significant challenges for security teams trying to detect misuse or investigate incidents."

The risks inherent in this behavior are manifold. When an employee inputs corporate data into a public-facing Large Language Model (LLM) or grants an AI agent broad permissions, that information may be stored on third-party servers that fall outside the organization’s security perimeter. Furthermore, there is the persistent risk that the data could be utilized to train future iterations of those models, potentially leading to the regurgitation of proprietary information in responses provided to other users or organizations. Additionally, autonomous agents themselves represent an expanding attack surface; if a malicious actor hijacks an AI agent that has been granted broad administrative privileges, the potential for data exfiltration or system disruption is profound.

Security Leaders Caught in a Visibility Gap

While employees are driving the adoption of shadow AI, the report suggests that IT and security executives are struggling to maintain visibility and control. The findings reveal a striking paradox: while security leaders express high levels of confidence in their oversight, that confidence is often misplaced.

A Fifth of UK Employees Share Passwords with AI

According to the study, 92% of IT and security executives claimed they were confident that they had complete visibility into every AI agent operating within their corporate environment. Yet, this assertion stands in direct contradiction to the self-reported behavior of the workforce. One-quarter (25%) of employees openly admitted to using AI tools that their employers are completely unaware of. This gap between executive perception and the reality on the ground indicates that many security teams are blind to a significant portion of the AI-driven activity occurring within their networks.

The consequences of this lack of visibility are already being felt. The Dashlane report found that 72% of the IT and security executives surveyed reported that their organization had experienced a security incident directly involving an AI agent’s credentials or access within the past 12 months. This figure serves as a sobering reminder that the risks associated with AI are not merely theoretical—they are a present and costly reality for businesses navigating the current technological landscape.

For many security leaders, this environment is creating a sense of professional vertigo. More than two-thirds (68%) of UK IT security executives expressed that they feel caught in a difficult tug-of-war: they are under immense pressure from the business to deploy AI technologies at speed to remain competitive, yet they are simultaneously tasked with maintaining rigorous security standards in an era where the traditional perimeter has effectively dissolved.

The pace at which AI is being integrated into the workplace is unprecedented. Unlike the slow adoption cycles of traditional software deployments, AI tools are often accessible via a web browser, making it trivial for an individual employee to onboard a tool and begin feeding it corporate data within seconds. This velocity makes it nearly impossible for legacy security tools, which were designed for more stable, managed environments, to keep up.

Frenet emphasized that the traditional security playbooks are no longer sufficient to handle the nuances of AI integration. "The challenge for security leaders is that the pace of change presented by AI and the potential consequences of misuse is unlike any other technology before it," he noted.

To mitigate these risks, the report suggests that organizations must return to the foundational principles of cybersecurity, but apply them with a new focus on non-human identities. "The principles that make up the foundation of a good security program for securing human identities—just-in-time provisioning, least privilege, good credential hygiene, and authentication—have to be applied to our new AI colleagues as well," Frenet concluded.

Ultimately, the findings from Dashlane underscore an urgent need for organizations to bridge the gap between their security policies and the daily reality of their employees. As AI continues to evolve, the distinction between a helpful productivity tool and a significant security liability becomes increasingly blurred. Without proactive management, clear governance frameworks, and better employee education, the "AI Colleague Effect" is likely to remain a significant blind spot for UK businesses, leaving them vulnerable to the next wave of sophisticated, AI-enabled threats. For now, the responsibility lies with leadership to transition from a culture of reactive, ad-hoc AI adoption to one of secure, governed, and transparent AI integration.

Source : Finance Poetry
Author : rifanmuazin
Editor : Finance Poetry Editorial Team