Artificial intelligence (AI) has truly “entered the building,” crossing the threshold from an emerging technology to an everyday business tool. The pace of adoption caught many organizations unprepared. While business leaders worked to formulate their AI strategy, employees discovered AI tools help complete routine work faster and more effectively, often before visibility, policies, and controls were established.
This unapproved or ungoverned use of AI, commonly called “shadow AI,” can expose sensitive data, introduce vulnerable software, create compliance issues, and leave organizations with little understanding of how AI is interacting with their wider environment.
For the most part, organizations are aware of the problem. 69% already suspect or have evidence that employees are using unsanctioned AI agents and tools, while 90% of IT leaders are concerned about shadow AI risk.
Awareness isn’t visibility, however. Many security and IT teams lack a deep understanding of that exposure, and cannot answer questions such as:
- Which AI tools and services are in use, and who is using them?
- What data, applications, and systems can those tools access?
- Where has that activity introduced security or compliance risk?
How big is the problem?
AI adoption, use, and development are advancing faster than many organizations can govern them. New tools and AI-enabled features can appear quickly, spread organically via employee adoption or routine software updates, and connect with sensitive systems and data. In many cases, shadow AI is a far bigger problem than your customers realize.
This is not just about ChatGPT
Shadow AI use goes beyond employees and business leaders using public chatbots such as ChatGPT or Gemini. Unauthorized or ungoverned AI can enter an organization via multiple routes:
| Category | Application of AI |
| Standalone AI tools | Public AI chatbots, assistants, and content generation tools |
| Non-enterprise versions of AI-enabled workplace applications | |
| Specialized AI tools for coding, research, design, and other business tasks | |
| AI embedded in existing software | Native OS and device assistants |
| Copilots embedded in specialist software applications | |
| AI capabilities added through automatic software updates or feature releases | |
| Browser and workspace tools | Third-party browser extensions |
| Workspace integrations and add-ons | |
| Email, productivity, and collaboration plugins | |
| Automation and connected AI | AI-enabled automation workflows |
| No-code/low-code AI integrations | |
| AI agents and third-party tools connected to business applications or company data |
Adoption speed is outpacing governance
AI capabilities are appearing across an expanding range of applications. Meanwhile, employees under pressure to improve productivity will explore AI-enabled tools to help them work faster.
Governance was already behind
Just one in 10,000 organizations has a CISO, leaving many businesses without dedicated executive-level security leadership to drive governance and risk assessment programs. AI increases the pressure on your customers’ constrained teams by accelerating the pace and scale of technology adoption they must govern.
The opportunity for partners
Organizations have spent years building processes to manage devices, applications, cloud workloads, and identities. Yet many still cannot confidently answer a basic question: How many AI tools are being used across their business?
Your customers have faced disruptive technology shifts before, from cloud adoption and remote work to the rapid evolution of ransomware. AI creates a new challenge, one many of your customers cannot answer: How do we securely adopt AI while managing unfamiliar and rapidly evolving risks? MSPs can deliver the security expertise, operational capacity, and ongoing guidance that many of your customers cannot achieve on their own.
Five questions to start the conversation
Partners that start these conversations today can reinforce their role as strategic advisors, not merely technology vendors. Instead of opening with product positioning, start by asking questions that help your customers understand their exposure:
- Which AI tools and services are employees using across your organization?
- What sensitive data, applications, and systems can those tools access?
- What policies and controls govern AI access and use?
- How do you identify and address AI-related security and compliance risks?
- Can leadership confidently explain the organization’s current AI risk posture?
For many customers, their answers will reveal gaps they didn't know existed and create a path for you to help them take the next step (or in many instances, the first step) in their AI security journey.
Visibility comes first
AI can deliver your customers productivity gains, faster decision-making, and competitive advantages. Realizing those benefits requires safe AI adoption: visibility into how AI is being used, controls that reduce inappropriate access and data exposure, and protection against AI-related threats.
Many of your customers know shadow AI is a concern. The problem is that many lack the tools, time, or expertise to tackle it on their own. As their trusted partner, you can lead the way. The first step is visibility: helping them uncover where and how AI is already in use across their organization.
Sophos AI Defense enables you to bring customers’ AI use out of the shadows and into view, so you can apply policies that govern access and data sharing. With this greater visibility and control, you can help customers reduce the risks of shadow AI while supporting productive, responsible adoption.


