Skip to content
All articles
Software & Access22 min read

Shadow IT Discovery: Finding and Managing Unsanctioned Software and Access

This article is general information and not legal advice; consult qualified counsel for legal interpretation specific to your organization.

Shadow IT—technology used without formal IT or security approval—often starts with good intentions: employees need to solve a problem quickly, collaborate with a vendor, automate a task, or try an artificial intelligence (AI) tool. The problem is that unapproved software can create unmanaged access, data exposure, compliance gaps, and incident response blind spots. In this article, you’ll learn how to build a practical shadow IT discovery program, evaluate employee AI tool usage risks, and connect findings to a repeatable user access review process.

What Shadow IT Really Means

Shadow IT includes any application, platform, browser extension, cloud service, AI tool, file-sharing site, software-as-a-service (SaaS) product, script, or device used for business purposes without appropriate visibility, approval, or governance.

Examples include:

  • A department paying for a SaaS project management tool with a corporate card
  • Employees using personal file-sharing accounts to exchange work documents
  • Teams adding browser extensions that can read page content or capture screenshots
  • Staff pasting company data into public AI chat tools
  • A manager creating an unofficial customer database in a cloud spreadsheet
  • Contractors accessing systems through shared credentials
  • Teams connecting third-party applications to Microsoft 365, Google Workspace, Slack, Salesforce, or other core platforms without security review

Shadow IT is not always malicious. In many cases, it reflects unmet business needs, slow procurement processes, limited approved tools, or lack of awareness. The goal is not to punish innovation. The goal is to find what is being used, understand risk, and bring useful tools under appropriate controls.

Why Shadow IT Discovery Matters

You cannot protect, monitor, back up, review, or retire systems you do not know exist. Shadow IT discovery helps organizations answer critical security and compliance questions:

  • What software is being used to store, process, or transmit business data?
  • Who has access to each application?
  • Are former employees, vendors, or contractors still active in unmanaged tools?
  • Are sensitive records being uploaded to platforms with unclear privacy or retention practices?
  • Are employees granting third-party apps access to email, files, calendars, or customer data?
  • Are AI tools being used with regulated, confidential, or proprietary information?
  • Are applications configured with multifactor authentication, role-based access, logging, and data retention controls?

For healthcare organizations and vendors supporting healthcare clients, these questions are especially important. If electronic protected health information (ePHI) is involved, the HIPAA Security Rule requires administrative, physical, and technical safeguards. Relevant provisions may include the security management process at 45 CFR 164.308(a)(1), workforce security at 45 CFR 164.308(a)(3), information access management at 45 CFR 164.308(a)(4), security awareness and training at 45 CFR 164.308(a)(5), access control at 45 CFR 164.312(a), audit controls at 45 CFR 164.312(b), integrity at 45 CFR 164.312(c), and transmission security at 45 CFR 164.312(e). Applicability depends on the organization’s role, data involved, and specific circumstances.

Shadow IT can also affect obligations under contracts, cyber insurance policies, state privacy laws, payment card requirements, and security frameworks such as the NIST Cybersecurity Framework, NIST Special Publication 800-53, CIS Critical Security Controls, ISO/IEC 27001, and SOC 2 Trust Services Criteria. Requirements vary by framework version, customer agreement, payer expectation, state law, and business context.

Common Sources of Shadow IT

Shadow IT usually appears in predictable places. Start discovery efforts where employees are most likely to adopt tools without formal approval.

Department-Owned SaaS

Marketing, sales, finance, human resources, operations, and clinical administration teams often subscribe to specialized tools. These may include:

  • Survey platforms
  • Design and content tools
  • Scheduling systems
  • Customer relationship management add-ons
  • Billing or reporting tools
  • Recruiting platforms
  • Training portals
  • Workflow automation platforms

The risk is not simply that these tools exist. The risk is that they may store sensitive data, lack centralized authentication, or remain active after users leave the organization.

Collaboration and File Sharing Tools

Employees frequently choose convenience when sharing files. Shadow collaboration tools may include:

  • Personal cloud storage accounts
  • Unapproved shared drives
  • External document collaboration spaces
  • Messaging applications
  • Video meeting platforms
  • Note-taking tools

These tools can become unmanaged repositories of contracts, patient information, financial records, credentials, screenshots, and internal discussions.

Browser Extensions and Connected Apps

Browser extensions and OAuth-connected applications can create significant exposure. OAuth is an authorization method that allows one application to access another without sharing a password. For example, a scheduling app may request permission to read a user’s calendar.

Some permissions are reasonable. Others are broad, such as access to email, files, contacts, or the ability to send messages on behalf of the user. Shadow IT discovery should include both installed browser extensions and third-party applications connected to core identity and productivity platforms.

Developer and Automation Tools

Technical teams may use code repositories, testing platforms, cloud services, scripts, application programming interface (API) tools, and automation services. These tools can hold credentials, tokens, source code, test data, or production data.

Security review should not block legitimate engineering work, but it should ensure secrets are protected, access is limited, logging is enabled, and sensitive data is not copied into unmanaged environments.

Employee AI Tool Usage

Employee AI tool usage risks deserve focused attention because AI tools are easy to access and often useful. Employees may use AI to draft emails, summarize documents, generate code, analyze spreadsheets, write policies, or troubleshoot technical issues.

Risks include:

  • Uploading regulated, confidential, or proprietary data into tools not approved for that data type
  • Exposing source code, credentials, customer information, patient details, or internal strategy
  • Relying on inaccurate AI-generated outputs without review
  • Creating records that are difficult to retrieve, monitor, or delete
  • Violating contractual confidentiality obligations
  • Using AI tools that retain prompts or outputs in ways inconsistent with company policy
  • Connecting AI tools to email, files, chat, ticketing, or customer systems with excessive permissions

Organizations should not assume employees are avoiding AI tools. A better approach is to provide clear rules, approved options, training, and monitoring aligned to business needs.

Build a Shadow IT Discovery Program

Shadow IT discovery works best as an ongoing program, not a one-time cleanup. The program should combine technical detection, financial review, access governance, employee engagement, and risk-based decision-making.

Step 1: Define What You Are Looking For

Create a clear internal definition of shadow IT. Include:

  • SaaS applications
  • Locally installed software
  • Mobile applications used for work
  • Browser extensions
  • AI tools
  • Cloud infrastructure accounts
  • Connected third-party apps
  • Automation platforms
  • File storage and collaboration tools
  • External portals containing company or regulated data
  • Shared accounts and unmanaged credentials

Define what counts as approved, conditionally approved, prohibited, and under review.

Example categories:

| Category | Meaning | Example Action | |---|---|---| | Approved | Reviewed and authorized for defined use | Add to software inventory and access review cycle | | Conditionally approved | Allowed only for certain data or users | Document restrictions and monitor usage | | Under review | Discovered but not yet assessed | Temporarily limit sensitive data use | | Prohibited | Not allowed due to risk or legal/contract concerns | Block or remove after business transition |

Step 2: Inventory Known Applications First

Before hunting for unknown tools, consolidate what you already know. Gather data from:

  • IT asset management systems
  • Endpoint management platforms
  • Identity provider applications
  • Single sign-on portals
  • Procurement and finance systems
  • Expense reports and corporate card statements
  • Vendor management lists
  • Contract repositories
  • Help desk tickets
  • Security questionnaires
  • Data flow diagrams
  • Departmental process documentation
  • Cloud administration consoles

Create a centralized software inventory that includes:

  • Application name
  • Business owner
  • Technical owner
  • Department
  • Purpose
  • Data types processed
  • User population
  • Authentication method
  • Multifactor authentication status
  • Vendor risk status
  • Contract status
  • Renewal date
  • Logging availability
  • Data retention requirements
  • Integration points
  • Approval status

This inventory becomes the foundation for shadow IT discovery, access reviews, vendor reviews, incident response, and compliance evidence.

Step 3: Use Technical Signals for Shadow IT Discovery

Technical discovery should be broad enough to identify meaningful use without becoming invasive or impractical. Use data sources already available in your environment where appropriate.

Useful signals may include:

  • DNS logs showing domains visited by corporate devices
  • Secure web gateway or firewall logs
  • Cloud access security broker (CASB) or security service edge (SSE) tools
  • Endpoint detection and response (EDR) software inventory
  • Mobile device management records
  • Browser management tools
  • Identity provider sign-in logs
  • OAuth consent logs
  • Email security logs
  • Data loss prevention alerts
  • Network proxy logs
  • Software installation reports
  • API gateway logs
  • Cloud infrastructure account inventories

Look for repeated patterns, not isolated web visits. One employee reading a vendor’s public webpage is different from 40 employees logging into an unapproved SaaS platform every day.

Step 4: Review Finance and Procurement Data

Many shadow IT tools are discovered through spending records. Review:

  • Corporate card transactions
  • Expense reimbursements
  • Departmental subscriptions
  • Small recurring charges
  • Software purchases below procurement thresholds
  • Consulting invoices that include embedded software
  • App marketplace charges
  • Cloud service charges

Partner with finance and procurement to flag software-like purchases. Train approvers to ask basic questions:

  • Is this software, a subscription, a cloud service, or an AI tool?
  • Will company, customer, patient, employee, or financial data be entered?
  • Will users log in with company credentials?
  • Does it integrate with any internal system?
  • Is there a contract, data processing agreement, business associate agreement if applicable, or security review?

For healthcare organizations, whether a business associate agreement is needed depends on the relationship and whether the vendor creates, receives, maintains, or transmits protected health information on behalf of a covered entity or business associate. Legal interpretation should be reviewed with qualified counsel.

Step 5: Ask Employees Without Creating Fear

If employees believe discovery is a punishment exercise, they will hide tools. If they understand the goal is safer enablement, they are more likely to cooperate.

Use a short survey to identify tools that technical logs may miss. Ask:

  • What tools do you use to perform your work?
  • Which tools are not provided by IT?
  • Do you use AI tools for drafting, summarizing, coding, analysis, or administrative work?
  • Do you store or share customer, patient, employee, financial, or confidential data in any tool?
  • Are any tools paid for through expense reimbursement or a department budget?
  • Are there approved tools that do not meet your needs?
  • Are there manual workflows you are trying to automate?

Pair the survey with a positive message: “We are identifying tools people rely on so we can support work securely, reduce duplicate spending, and provide better approved options.”

Step 6: Prioritize by Risk

Not every shadow IT finding needs the same response. Prioritize based on data sensitivity, access scope, business criticality, and technical control gaps.

High-risk indicators include:

  • ePHI, personally identifiable information, payment data, credentials, legal records, or confidential business data
  • No multifactor authentication
  • Shared accounts
  • Former users still active
  • Public file links
  • Broad third-party app permissions
  • Unknown data retention or deletion practices
  • No contract or security review
  • Integrations with core systems
  • Administrative access held by a single employee
  • No audit logs
  • Use by many employees across departments
  • AI tools receiving sensitive prompts or files
  • Vendor located or processing data in jurisdictions that require further review under organizational policy or applicable law

Lower-risk findings may include limited use of a tool with no sensitive data and no integration to company systems. Even then, the tool should be documented and either approved, restricted, replaced, or retired.

Managing Employee AI Tool Usage Risks

AI governance should be integrated into shadow IT discovery rather than treated as a separate issue. Employees often view AI tools as websites, not software systems. Security teams should make the policy simple enough to follow.

Create an AI Use Classification Model

Define what employees may and may not enter into AI tools.

Example:

| Data Type | Public AI Tool | Approved Enterprise AI Tool | Notes | |---|---:|---:|---| | Public information | Allowed | Allowed | Still review output accuracy | | Internal non-sensitive information | May be allowed by policy | Allowed | Avoid confidential strategy or credentials | | Confidential business data | Usually restricted | Allowed only if approved | Requires contractual and security review | | ePHI or regulated data | Prohibited unless specifically approved | Only if approved for that use | Requires privacy, security, and legal review | | Credentials, secrets, tokens | Prohibited | Prohibited | Use approved secrets management processes | | Source code | Restricted | Depends on policy | Review intellectual property and security risks |

Your classifications should reflect your legal obligations, customer commitments, contracts, risk tolerance, and technical controls.

Provide Approved AI Paths

If employees need AI capabilities, offer approved options. This may include enterprise AI tools with administrative controls, logging, data protection terms, identity integration, and access management. Define approved use cases, such as:

  • Drafting general communications
  • Summarizing non-sensitive meeting notes
  • Creating training outlines
  • Assisting with code review under defined conditions
  • Generating non-sensitive spreadsheet formulas
  • Rewriting internal documentation

Also define prohibited or restricted use cases, such as:

  • Entering patient information into unapproved AI tools
  • Uploading contracts under negotiation without approval
  • Using AI-generated clinical, legal, financial, or security decisions without qualified human review
  • Pasting credentials, private keys, tokens, or vulnerability details into public tools
  • Connecting AI tools to email or file repositories without review

Monitor and Educate

Monitoring should align with company policy, employment obligations, and applicable law. Useful controls may include:

  • Web filtering for prohibited AI platforms
  • Alerts for uploads to unapproved AI domains
  • Data loss prevention rules for sensitive identifiers
  • Browser controls for extensions
  • OAuth app review
  • Approved AI tool logging
  • Employee training with real examples

Training should be direct and practical. For example:

> “Do not paste patient information, employee records, customer confidential data, passwords, access tokens, or non-public financial information into unapproved AI tools. If you need AI support for a sensitive use case, contact IT or security so we can identify an approved option.”

Connect Shadow IT to the User Access Review Process

Discovery is only half the problem. Once an application is found, you need to know who has access and whether that access is appropriate. A user access review process is a structured review of user accounts and permissions to confirm that access is still needed, correctly assigned, and promptly removed when no longer appropriate.

Why Access Reviews Matter

Shadow IT often bypasses normal joiner-mover-leaver controls:

  • New hires are added manually by a department administrator
  • Employees who change roles keep old access
  • Former employees remain active because the tool is not tied to single sign-on
  • Contractors are not removed at contract end
  • Shared accounts hide individual accountability
  • Administrative privileges are granted without review

A good access review process reduces these risks and supports compliance expectations in many frameworks.

Build a Practical Access Review Cycle

For each approved or discovered application, define:

  • Application owner
  • Access reviewer
  • Review frequency
  • User list source
  • Role or permission descriptions
  • Criteria for approval
  • Evidence to retain
  • Remediation timeline
  • Escalation path

Suggested review frequencies should be risk-based. For example:

  • Critical systems and systems containing regulated or sensitive data: more frequent reviews
  • Administrative accounts: more frequent reviews
  • Low-risk tools with limited data: less frequent reviews
  • Recently discovered shadow IT: initial review immediately, then risk-based schedule

Avoid setting a universal review frequency unless it fits your organization’s obligations and risk profile. Requirements may vary by framework, customer contract, payer expectation, or regulator.

Access Review Checklist

Use this checklist during reviews:

  • [ ] Is each user still employed or under active contract?
  • [ ] Does each user still need access for their current role?
  • [ ] Are permissions aligned to least privilege?
  • [ ] Are administrative roles limited and justified?
  • [ ] Are shared accounts removed or formally exception-approved?
  • [ ] Is multifactor authentication enabled where appropriate?
  • [ ] Are inactive accounts disabled or removed?
  • [ ] Are external users still authorized?
  • [ ] Are service accounts documented with owners and purpose?
  • [ ] Are integrations and API tokens still needed?
  • [ ] Are OAuth-connected apps reviewed?
  • [ ] Are access changes documented?
  • [ ] Are removals verified after the review?
  • [ ] Is evidence retained for audit or compliance needs?

Example Access Review Workflow

  1. Export user list from the application or identity provider.
  2. Identify owner responsible for validating business need.
  3. Compare users against HR, contractor, and vendor records.
  4. Flag anomalies, including terminated users, inactive accounts, excessive permissions, shared accounts, and unknown external users.
  5. Send review to owner with clear approval or removal options.
  6. Apply changes within a defined remediation window.
  7. Validate completion by re-exporting user lists or checking logs.
  8. Retain evidence, including reviewer name, date, decisions, and completed changes.
  9. Update inventory with review date, owner, and risk status.

Keep the process simple enough that business owners can complete it accurately. If reviewers do not understand permission names, translate them into business terms.

Decide: Approve, Replace, Restrict, or Retire

Once shadow IT is discovered and access is reviewed, decide what to do with it.

Approve

Approval may be appropriate when the tool solves a legitimate business need and can be governed effectively.

Approval actions:

  • Complete vendor risk review
  • Confirm contract and privacy terms
  • Determine whether a business associate agreement or data processing agreement is needed
  • Enable single sign-on where possible
  • Require multifactor authentication
  • Assign business and technical owners
  • Configure logging
  • Define allowed data types
  • Add to access review process
  • Document retention and deletion expectations
  • Add to incident response contacts and playbooks

Replace

Replacement may be best when an approved tool already provides the same function or when the shadow tool cannot meet security needs.

Replacement actions:

  • Identify business requirements
  • Map data migration needs
  • Set transition timeline
  • Communicate approved alternative
  • Disable new use of the shadow tool
  • Export or archive business records if needed
  • Delete data from the old platform according to policy and contract
  • Remove accounts and revoke integrations

Restrict

Some tools may be allowed only for limited use.

Restriction examples:

  • Public information only
  • No ePHI or regulated data
  • No customer uploads
  • No integrations with email or file repositories
  • Only specific departments
  • No administrative access outside IT
  • Read-only access for external users
  • Approved AI prompts only

Restrictions must be easy to understand and enforce. A policy that employees cannot interpret will not work.

Retire

Retirement is appropriate when risk is unacceptable, use is unnecessary, or a tool duplicates approved services.

Retirement actions:

  • Notify users and owners
  • Preserve required records
  • Export business data
  • Revoke user access
  • Disable integrations and tokens
  • Cancel subscriptions
  • Request data deletion where appropriate
  • Validate account closure
  • Update inventory and financial records
  • Monitor for reappearance

Create Policies That Employees Can Follow

Policies should support secure work, not just prohibit behavior. Consider creating or updating:

  • Acceptable use policy
  • Software acquisition policy
  • AI acceptable use policy
  • Access control policy
  • Vendor risk management policy
  • Data classification policy
  • Records retention policy
  • Mobile device and browser extension standards
  • Third-party integration policy

Keep policy language practical. For example:

Instead of: “Users shall not use unauthorized applications.” Use: “Before using a new software tool, browser extension, AI platform, file-sharing service, or automation application for company work, submit it through the software review process. Do not enter patient, customer, employee, financial, credential, or confidential business information into tools that have not been approved for that data type.”

Build a Lightweight Intake Process

Employees often use shadow IT because official processes are too slow. Create a fast intake path for software requests.

Minimum intake questions:

  • What problem are you trying to solve?
  • What tool do you want to use?
  • Who will use it?
  • What data will be entered, uploaded, stored, or transmitted?
  • Does it involve patient, customer, employee, financial, or confidential information?
  • Does it use AI features?
  • Does it integrate with any company system?
  • Is it free, paid, or trial-based?
  • When do you need it?
  • Is there an approved tool that might meet the need?

Then triage requests:

  • Low risk: quick approval with standard conditions
  • Medium risk: security and privacy review
  • High risk: formal vendor risk, legal, privacy, and technical review
  • Prohibited: explain why and offer alternatives where possible

Publish expected turnaround times. If every request disappears into a queue, employees will work around the process.

Technical Controls That Help Reduce Shadow IT

Policies and training are necessary, but technical controls make governance scalable.

Consider these controls based on your environment and risk profile:

  • Single sign-on for approved applications
  • Multifactor authentication
  • Centralized identity lifecycle management
  • Endpoint software inventory
  • Browser extension allowlisting or blocklisting
  • OAuth app approval workflows
  • DNS and web filtering
  • CASB or SSE visibility
  • Data loss prevention
  • Cloud storage controls
  • Mobile application management
  • Privileged access management
  • SaaS security posture management
  • Centralized logging and alerting
  • Automated deprovisioning tied to HR records
  • Procurement system flags for software purchases

Avoid relying on blocking alone. If users have a real business need, they will seek another workaround. Pair controls with approved alternatives.

Metrics to Track

Measure progress with practical indicators:

  • Number of discovered unapproved applications
  • Number of applications classified by risk
  • Percentage of applications with assigned owners
  • Percentage of applications with completed access reviews
  • Number of inactive or terminated users removed
  • Number of applications using single sign-on
  • Number of applications with multifactor authentication enabled
  • Number of high-risk OAuth apps removed or approved
  • Number of AI tools discovered
  • Number of users trained on AI and software usage rules
  • Time to review new software requests
  • Duplicate tools retired
  • Recurring shadow IT spending reduced or consolidated

Use metrics to improve the program, not to shame departments. Trends matter more than one-time counts.

Example 30-60-90 Day Plan

First 30 Days: Establish Visibility

  • Define shadow IT and approval categories
  • Consolidate known software inventory
  • Pull identity provider application lists
  • Review corporate card and expense data for software subscriptions
  • Identify top unapproved SaaS and AI tools from logs
  • Launch a short employee survey
  • Prioritize tools handling sensitive or regulated data
  • Draft interim AI usage guidance

Days 31-60: Reduce Highest Risks

  • Assign owners for high-risk applications
  • Conduct access reviews for top-risk tools
  • Remove terminated, inactive, and unnecessary users
  • Revoke risky OAuth permissions
  • Disable shared accounts where feasible
  • Review AI tools receiving sensitive data
  • Define approved and prohibited AI use cases
  • Start vendor risk reviews for tools likely to remain
  • Communicate approved alternatives

Days 61-90: Operationalize

  • Finalize software intake workflow
  • Add approved tools to access review schedule
  • Integrate procurement and finance alerts
  • Implement browser extension and OAuth governance
  • Update policies and training
  • Retire or replace unnecessary tools
  • Establish recurring reporting to leadership
  • Document evidence for compliance and audits

Leadership Questions to Ask

Business and technology leaders should regularly ask:

  • Do we know which applications are used across the organization?
  • Which tools store regulated, confidential, or business-critical data?
  • Are employees using AI tools, and under what rules?
  • Can we remove access promptly when employees leave?
  • Do business owners review access regularly?
  • Are departments buying software outside procurement?
  • Are there approved alternatives for common shadow IT use cases?
  • Are we blocking risky tools without addressing the underlying business need?
  • Do our incident response plans include SaaS and AI tools?
  • Can we produce evidence of access reviews and remediation?

These questions help keep shadow IT management connected to business risk rather than treating it as a purely technical cleanup effort.

Conclusion and Clear Next Steps

Shadow IT discovery is not about stopping employees from solving problems. It is about gaining visibility, reducing unmanaged risk, and giving teams secure ways to work. Start by building a reliable software inventory, combining technical and financial discovery, addressing employee AI tool usage risks, and connecting every meaningful application to a user access review process.

Next steps:

  1. Define what counts as shadow IT in your organization.
  2. Inventory known applications and owners.
  3. Review logs, expense data, and employee feedback for unknown tools.
  4. Prioritize applications by data sensitivity and access risk.
  5. Review users and permissions for high-risk tools first.
  6. Publish clear AI and software usage rules.
  7. Create a fast intake process for new tools.
  8. Track remediation and repeat discovery on a regular schedule.

How ThornShield Can Help

ThornShield Technologies helps organizations identify software usage, assess access risk, and build practical governance processes for SaaS, AI tools, and unmanaged applications. Our Software Usage & Access Analysis service can help you find shadow IT, review user access, and create a prioritized remediation plan aligned to your business and compliance needs. This article is general information and not legal advice; consult qualified counsel for legal interpretation specific to your organization.

Find out where you stand.

Tell us about your organization and what worries you most. We'll come back with an honest view of your risks and the most practical way to address them.