Shadow AI: 8 Places It Hides

Banning AI does not stop people using it — it only stops you seeing it. The eight places it hides, how to find each one, and the architecture that turns shadow AI into governed AI.

8places shadow AI hides
6distinct risks it creates
5steps to bring it into the light
1rule: offer a better approved path than the shadow one

Shadow AI is any AI tool, model, feature or integration used for work without formal approval or oversight. It is the successor to shadow IT, with one important difference: shadow IT stored your data somewhere you did not know about, whereas shadow AI reads your data, may retain it, and increasingly acts on your systems. The six risks follow from that — loss of visibility, data and privacy leakage, regulatory exposure, a wider attack surface, unreliable or biased output, and tool sprawl with vendor lock-in.

1 · The Eight Hiding Places

Roughly in order of how visible they are to a security team — the first is easy to spot, the last is nearly invisible.

#Where it hidesTypical useMain risk
01Employee browsers — public chatbots, AI browser extensionsDrafting emails, summarising documents, analysing pasted dataSensitive text leaves approved systems; extensions can read every page
02Personal accounts — privately registered AI accounts used for workContent, research, code, analysisNo enterprise contract, no retention terms, no offboarding
03SaaS applications — AI features switched on inside approved toolsMeeting notes, recommendations, document generationA new data processor arrives through an update and skips vendor review
04Department tools — AI platforms bought directly by a teamMarketing, sales intelligence, customer analyticsDuplicate spend, unreviewed contracts, compliance gaps
05Developer environments — unapproved models, APIs, coding assistants, librariesCode completion, tests, refactoringProprietary code and secrets reach external providers
06Workflow integrations — AI wired in through APIs, plugins, automation platformsRecord updates, data transfers, routingOver-broad OAuth scopes give an AI service standing access
07Uploaded documents — enterprise files pushed into external AIContract review, translation, extractionConfidential files stored or reused outside your control
08Embedded AI agents — third-party agents operating inside your applicationsTask execution, orchestration, support automationSoftware that acts with no owner, no monitoring and no approval

▸ Places 06 and 08 are where shadow AI stops being a data-leak problem and becomes an access-control problem: the AI is no longer just reading what someone pastes, it holds credentials and takes actions.

2 · How to Find Each One

Every hiding place leaves a trace in a system you probably already run. Discovery is mostly about reading logs you already collect with a new question in mind.

Hiding placeWhere the evidence isControl once found
BrowsersSecure web gateway / DNS logs for AI domains; managed-browser extension inventoryAllow-list approved AI; block unvetted extensions; DLP on paste/upload
Personal accountsGateway logs showing consumer (non-SSO) logins to AI servicesEnterprise tenant with SSO; tenant restrictions to block personal sign-in
SaaS featuresSaaS admin consoles, release notes, SaaS posture managementDefault AI features off until reviewed; add AI clauses to vendor review
Department toolsExpense claims, card spend, procurement recordsAI purchases routed through one intake and risk review
Developer environmentsRepository scanning for AI SDK imports and API keys; egress logs to model APIsApproved assistant with no-training terms; secret scanning; egress via the AI gateway
IntegrationsIdentity provider OAuth-grant and app-consent reportsAdmin consent for AI apps; least-privilege scopes; periodic grant review
Uploaded documentsDLP and gateway logs of file uploads to AI domainsClassification labels that block upload of confidential files
Embedded agentsService-account and API-key inventories; unusual machine identitiesEach agent gets its own identity, an owner, scoped tools and an audit trail
  1. Discover & inventory every AI tool, feature, model and agent in use — an AI register, like an asset register.
  2. Set a usage policy — which data classes may go to which tools, and who approves new ones.
  3. Provide an approved alternative that is at least as good as the shadow tool. This step decides whether the others work.
  4. Educate — explain the “why”: retention, training on inputs, prompt injection, wrong answers.
  5. Monitor, audit and enforce continuously — new AI features ship weekly; an annual review is already out of date.

3 · What Governed AI Looks Like

The approved path is an architecture, not a policy document. Every request — from a person or from an agent — passes the same checkpoints, in this order:

User / agent→Identity provider→API gateway / WAF→AI gateway→Guardrails & DLP→Authorisation→Model, RAG & tools→Enterprise systems
CheckpointWhat it doesShadow-AI risk it closes
Identity providerSSO, MFA, conditional access for people; managed identities for agentsPersonal accounts; agents with no owner
API gateway / WAFTLS, rate limiting, IP filtering, bot protectionAbuse and cost blow-outs
AI gatewayOne front door to approved models: routing, model allow-list, caching, usage and cost meteringUnapproved models and APIs; no visibility
Guardrails & DLPDetect prompt injection and jailbreaks; find and mask personal data and secrets; filter unsafe outputData leakage; manipulated output
AuthorisationRBAC/ABAC per user and per agent; token validation; least privilege on every tool callOver-privileged integrations and agents
RAG & vector storeRetrieval respects the caller’s permissions; embeddings encrypted at restAI that reveals documents the user could not open
ObservabilityCentral logs and audit trails of prompts, tool calls and decisions, sent to the SIEMActions nobody can reconstruct
Human in the loopApproval step before high-risk actions (payments, deletions, external messages)Autonomous agents acting without approval

▸ The zero-trust rules apply unchanged: verify every request, give every identity least privilege, assume breach. An AI agent is just another identity — one that works faster and never gets tired.

4 · The Frameworks to Anchor It

FrameworkWhat it gives you
ISO/IEC 42001:2023A certifiable AI management system — the AI equivalent of an ISO 27001 ISMS, including an AI inventory and impact assessments
NIST AI RMF 1.0 (2023)Four functions — Govern, Map, Measure, Manage — for AI risk, free to use
OWASP Top 10 for LLM ApplicationsThe technical risk list: prompt injection, sensitive-information disclosure, excessive agency and more
India’s DPDP Act 2023Pasting customers’ personal data into an external AI is processing; the organisation stays answerable as data fiduciary
EU AI Act (Regulation 2024/1689)Risk-tiered obligations for providers and deployers of AI used in or sold into the EU

For the attacks themselves, see AI Security: New Attack Surface; for how the models work underneath, How LLMs Work, Explained.

Bottom line

  • Banning drives it underground — the only durable fix is an approved path that is easier than the shadow one.
  • Eight places, eight traces — gateway, DLP, OAuth, procurement and repository logs already hold the evidence.
  • Agents turn a leak into an access problem — give every agent an identity, an owner and scoped tools.
  • One front door — an AI gateway with guardrails makes every model call visible, governed and billable.
  • Inventory first — ISO 42001 and the NIST AI RMF both start with knowing what AI you run.

Sources & method. Hiding places and risks reflect common enterprise AI-governance practice; detection and control mappings are the author’s analysis. Framework scopes from ISO/IEC 42001:2023, NIST AI 100-1 (AI RMF 1.0), the OWASP Top 10 for LLM Applications, the Digital Personal Data Protection Act 2023 and Regulation (EU) 2024/1689. Product-neutral: no vendor is implied by any checkpoint. Original analysis; no third-party graphics or text reproduced.

← Back to Cyber security