Quick answer
Shadow AI is the use of AI tools, models or features at work without approval or oversight from IT and security. It includes staff pasting company data into public chatbots, AI features switched on inside existing software, and teams building AI integrations without review.
Why It Happens
AI tools are free or cheap, useful and one browser tab away. When organisations have not provided approved tools or clear guidance, staff use whatever helps them get work done. Many software vendors have also added AI features to products already in use, sometimes enabled by default.
The Risks
- Data exposure: confidential, personal or client data entered into tools that may store it, use it for training or be breached.
- No access control: AI integrations connected to email, drives or business systems without security review.
- Unvetted output: AI-generated code, advice or documents used without checks.
- Compliance gaps: personal information processed by unknown third parties.
How to Find It
- Review network and proxy logs for traffic to AI services.
- Audit OAuth and app consents in your identity platform for AI applications with mailbox or file access.
- Check which AI features are enabled in existing software.
- Ask staff, without blame, which tools they use and why.
How to Govern It
Provide an approved AI tool so staff have a safe option, publish a short acceptable use policy for AI, block or restrict high-risk services, review AI integrations before connection and test approved AI systems before broad rollout. An AI security assessment shows what your approved tools could expose.
Regulatory and Framework Context
For Australian organisations, shadow AI creates obligations under the Privacy Act wherever personal information is entered into AI tools.
Latest intel on this topic
- Match fined $10 million for alleged Tinder privacy violationsThe Record · 16 November 2026
- Italy Fines Meta €805 Million Over Facebook's Android Gallery Access Without Explicit ConsentThe Hacker News · 15 November 2026
- Record number of cyber incidents reported to UK privacy regulatorThe Record · 13 November 2026
- The hidden risks of shadow AIUK NCSC · 7 September 2026
Updated hourly from the daily briefing.
Common questions
Should we block ChatGPT and other public AI tools?
Blocking alone usually pushes use onto personal devices. Providing an approved tool with enterprise data protections, plus clear policy, works better.
What is an AI acceptable use policy?
A short policy setting out which AI tools staff may use, what data may be entered into them and how AI output must be checked.
Change log
- 13 November 2026First published.