# Shadow AI

Shadow AI is the use of AI tools, models, or capabilities for organizational work without appropriate approval, visibility, or governance.

Cloud concept. Updated 2026-10-29. Page: https://blowstack.com/tools/cloud-concepts-hub/shadow-ai

> Shadow AI occurs when employees or teams adopt AI outside established security, procurement, and risk-management processes, including AI features embedded in otherwise familiar software. Transparent classification of tools as approved, blocked, or under evaluation helps people choose permitted solutions while reducing data exposure, compliance gaps, and unaccountable decision-making.

Practice exams covering Shadow AI: [AIB-C01 (beta) AWS Certified AI Business Strategist](https://blowstack.com/practice-exams/aws-certified-ai-business-strategist).

## 6 key aspects of Shadow AI

Reducing shadow AI requires visible classifications, scoped approvals, controlled experimentation, layered enforcement, and usable alternatives that remain under active review.

1. **Transparent tool classification**: Publish approved, blocked, and under-evaluation statuses in a catalog employees can easily find. Explain each status, its scope, and how to request a review.
2. **Use-case-specific approval**: Associate approval with permitted tasks, data classifications, editions, and configurations. Approval for public-content drafting must not be interpreted as approval to upload confidential records.
3. **Controlled evaluation**: Give evaluations an accountable owner, limited participants, suitable test data, and an expiration date. Keep production use prohibited unless explicitly authorized.
4. **Layered enforcement and visibility**: Combine identity permissions, cloud-account policies, network restrictions, and relevant endpoint or SaaS controls. Document coverage gaps, especially personal accounts and traffic that bypasses managed networks.
5. **Practical approved alternatives**: Provide sanctioned tools that meet real engineering and business needs. Straightforward onboarding and timely reviews reduce incentives to bypass governance.
6. **Continuous ownership and review**: Assign owners to tool records and reassess material changes in features, contracts, data handling, and integrations. Track exceptions and investigate discovered use without assuming every unauthorized user acted maliciously.

## Cloud services for Shadow AI

An approved application built with Amazon Bedrock can give employees a governed alternative to unsanctioned AI tools. IAM controls application and workload permissions, while AWS Organizations establishes account-level permission boundaries. CloudTrail supplies evidence of supported API activity, and AWS Network Firewall can restrict known external destinations on inspected network paths. These services support enforcement and investigation, but none replaces a published approval catalog or a risk-review process. Extend coverage with endpoint, browser, identity, and SaaS controls where work occurs outside governed AWS accounts and networks.

- **[Amazon Bedrock](https://blowstack.com/tools/cloud-services-hub/amazon-bedrock)**: Amazon Bedrock can provide a managed foundation for approved generative AI applications with IAM-controlled access and configurable safeguards. Using Bedrock does not automatically approve an application; its data flows, model choice, permissions, and intended use still require review.
- **[AWS Organizations](https://blowstack.com/tools/cloud-services-hub/aws-organizations)**: Service control policies can set permission boundaries for member accounts, such as denying selected AI service actions where they are not authorized. They do not grant permissions, maintain the tool-approval catalog, or govern external SaaS use.
- **[AWS CloudTrail](https://blowstack.com/tools/cloud-services-hub/aws-cloudtrail)**: CloudTrail records supported AWS API activity to help investigate AI service usage in governed accounts. Configure the required event coverage and use separate service-specific logging where needed; CloudTrail is not a universal record of prompts or external chatbot activity.
- **[AWS Network Firewall](https://blowstack.com/tools/cloud-services-hub/aws-network-firewall)**: AWS Network Firewall can apply domain-based filtering rules to supported HTTP and HTTPS traffic routed through it, helping restrict known unapproved AI endpoints. Its effectiveness depends on routing and protocol visibility, and it does not cover personal-device traffic outside that path.

## Shadow AI exam questions

Real practice exam questions on Shadow AI, with the correct answer and its explanation.

### Question 1 (AIB-C01 (beta))

**Question:**

An engineering company publishes a register that classifies AI tools as approved, blocked, or under evaluation. Developers use company-managed AWS roles to access approved AWS AI services. A review finds that some developers also upload company information to blocked public AI websites by using personal accounts on company-managed laptops. Leadership wants technical controls that enforce the register for both access paths without disrupting approved AWS work. Which approach best meets these requirements?

**Correct answer:** Use IAM policies to permit only authorized AWS AI operations, and use managed browser or web gateway controls to restrict access to blocked public AI tools. Keep these controls aligned with the published register.

**Explanation:** IAM can enforce permissions for AWS API access through company-managed roles, but it does not control personal accounts on external AI websites. Browser or web gateway controls address that separate access path. Aligning both controls with the register connects transparent classifications to enforceable restrictions.

## Practice Shadow AI in a full exam

Questions on Shadow AI in each practice exam, with every answer option and an explanation for each:

- [AIB-C01 (beta)](https://blowstack.com/practice-exams/aws-certified-ai-business-strategist): 8 questions

## What is Shadow AI

Shadow AI is an organizational governance problem, not a particular AI architecture or cloud service. It includes an employee pasting customer records into an unapproved chatbot, a developer connecting an application to a model API using a personal account, or a team enabling an AI meeting assistant without reviewing its recording and retention practices. The defining issue is missing authorization, oversight, or controls, rather than whether the technology is technically sophisticated or publicly accessible.

A practical starting point is a discoverable inventory that classifies AI tools as approved, blocked, or under evaluation. Each entry should identify the tool, accountable owner, permitted purposes, allowed data classifications, access requirements, restrictions, and review date. Classification should distinguish relevant editions, deployment configurations, and capabilities: an enterprise subscription with contractual protections may have a different status from the same vendor's consumer offering.

Approved means the tool may be used within documented boundaries, not that every use is safe. For example, an assistant might be approved for drafting public marketing copy but prohibited from processing customer identifiers. Blocked means its use is prohibited within the stated policy scope, with an explanation and an approved alternative where possible. Under evaluation means a decision is pending; it does not imply permission for unrestricted production use. Any evaluation access should have explicit limits, an owner, an expiration date, and suitable test data.

Cloud implementation combines this classification with technical enforcement. Organizations can provide approved model access through centrally managed accounts or an internal application, authenticate users through enterprise identity, and restrict permissions to authorized workloads. Organization-level policies can constrain AWS activity in governed accounts, while network controls can restrict access to known external AI endpoints on traffic paths they inspect. Endpoint, browser, and SaaS controls may also be necessary because cloud-account controls do not govern employees using personal accounts or devices.

Tool evaluation should examine data handling, retention, use of inputs for model training, access controls, contractual terms, regional processing, auditability, and suitability for the intended task. Solution selection also matters: a deterministic rules engine or conventional analytics workflow may be more appropriate than generative AI for a tightly specified calculation. A tool's approved status does not establish the correctness of its answers or authorize autonomous high-impact decisions.

For example, engineers who need help debugging should receive an approved coding assistant and clear rules about source code, credentials, and production logs. If a new browser extension offers similar assistance, its status can remain under evaluation while reviewers assess permissions and data transfers using synthetic examples. Publishing that status prevents employees from interpreting the absence of a decision as approval.

Blanket bans can reduce some exposure but also encourage concealment when useful alternatives are unavailable. Conversely, unrestricted experimentation can leak confidential information and create unsupported business dependencies. Effective governance balances usable approved tools, fast review processes, proportionate monitoring, and enforceable restrictions. Reassess classifications when features, contracts, integrations, or business uses change. Measure both adoption of approved tools and unresolved exceptions; a published catalog alone does not eliminate shadow AI.

## Learn Shadow AI in lessons

1 short lesson on Shadow AI in 1 certification course.

- AIB-C01 (beta): Shadow AI: transparent tool classification (Premium), https://app.blowstack.com/lessons/shadow-ai-transparent-tool-classification-aib-c01
