What is shadow AI?
Shadow AI is the use of AI tools inside an organisation without the knowledge or approval of IT and security teams. Typical examples include employees pasting company data into personal chatbot accounts, unvetted AI browser extensions, and AI features wired into business workflows outside any oversight.
Why shadow AI happens
It happens because the tools genuinely work. When a free chatbot saves an hour on a report and the sanctioned alternative does not exist, is locked down or is worse, employees make the rational choice. Shadow AI is rarely malicious; it is unmet demand. That is also why prohibition alone does not fix it. Blocking a domain usually pushes people toward a workaround rather than making them stop.
The risks it creates
Consumer AI accounts sit outside your contracts. Depending on the terms of service, submitted content may be retained or used to improve the vendor’s systems, and none of it appears in your audit trail. The consequences: data-leakage exposure for confidential and personal information, GDPR and sector-compliance gaps, and untracked AI output flowing into client deliverables. When an incident occurs, the organisation discovers it has no inventory of where AI touched the work.
How to respond
Treat shadow AI as a demand signal. Provide a sanctioned, enterprise-grade assistant with the data protections your policies require, so the right path is also the convenient one. Publish a short policy that names approved tools and the data rules for each. Train staff on what may never leave the building. Then monitor through SaaS discovery, network telemetry and expense reports, and onboard what you find instead of only punishing it.
How widespread is shadow AI?
Surveys across industries consistently find that a large share of employees, often a majority, use AI tools their employer has not approved, frequently with work data. If you have not measured it, the safe assumption is that it is already happening in your organisation.
Is blocking AI websites an effective fix?
On its own, no. Blocking moves usage to phones, personal laptops and lesser-known tools, which is worse for visibility. Blocking specific high-risk services can be part of the answer, but only alongside a sanctioned alternative people actually want to use.
What is the first step to getting it under control?
Discovery, then an offer. Measure actual usage for a few weeks, publish a one-page policy, and roll out an approved enterprise tool. Most shadow usage migrates voluntarily once a legitimate option exists.
- AI governanceAI governance is the set of policies, roles, processes and technical controls an organisation puts in place so that AI is used safely, legally and accountably. It defines which tools are approved, who may use them with which data, how usage is monitored, and how risks are assessed before a use case goes to production.
- EU AI ActThe EU AI Act (Regulation (EU) 2024/1689) is the world’s first comprehensive law regulating artificial intelligence. It entered into force in August 2024 and applies in stages. The Act classifies AI systems by risk level, from prohibited practices to strictly regulated high-risk systems, with lighter transparency duties for uses such as chatbots.
- Prompt injectionPrompt injection is an attack that hides malicious instructions inside content an AI system processes, such as an email, a document or a web page. The model then treats the attacker’s text as instructions and ignores the rules it was given. The OWASP Top 10 for LLM applications ranks it as the leading security risk for this type of system.
Deploy AI with confidence
Code75 implements production AI across enterprise teams, with the security testing and governance to match. You will talk to an engineer.