Staff use
Shadow AI: how staff use of AI tools leaks data
Shadow AI is AI use your organisation has not approved or does not control. It is a common route for data to leave, because the leak is a helpful employee, not an attacker.
Last reviewed against legislation.gov.uk, National Cyber Security Centre and OWASP GenAI Security Project. Sources
The problem
Helpful, quick and unlogged
Staff paste text into public AI tools to summarise, translate or debug. The data then sits on someone else's servers. Samsung's memo said data sent to AI platforms is stored on external servers, is hard to retrieve or delete and could become accessible to other users [8].
A real case
Samsung, April and May 2023
Reporting said engineers accidentally leaked internal source code by uploading it to ChatGPT in April 2023 [8]. Samsung then restricted generative AI on company devices and networks, and told staff using such tools on personal devices not to upload company data or intellectual property [8]. Samsung described the restriction as temporary and said it was reviewing safer ways to use the tools [8].
What the case does not tell us
The reporting does not say what the code was or whether anyone outside Samsung saw it. Treat it as an example of the route, not as evidence of harm.
Where it shows up
Four common places
Chatbots in a browser
Public chatbots need no installation or approval, so IT may never see them.
Browser extensions and plug-ins
Extensions that summarise pages or write emails may send the content elsewhere. Ask what each one does with it.
Features added to existing software
AI features can appear inside tools you already use. Check the supplier's data terms when they do.
Personal accounts on work devices
Work data entered through a personal account sits outside any agreement you have with the supplier.
What to do
Reduce it in order
-
1
List the AI tools in use by asking teams, not just by checking IT records. -
2
Decide which tools are approved and for what data. -
3
Write the rule and explain the reason. OWASP advises teaching users not to submit sensitive information [4]. -
4
Offer an approved alternative so staff are not left to improvise. -
5
Review the list when suppliers add AI features.
For the written rule see the AI use policy guide.
Keep reading
Read next
Exposed databases
What Wiz found in an open DeepSeek database, why it matters for any AI service and the checks UK teams can make on their own AI data stores.
Read the guide
Prompt leaks
How an LLM can disclose personal or confidential data through its outputs, what OWASP and the NCSC advise, and the design choices that limit it.
Read the guide
Reporting
When an AI data breach must be reported under UK GDPR, how the 72 hour clock works, and what to record whether or not you report.
Read the guideHow CyPro can help
Need a hand working through this?
Finding out whether an incident reaches you is one job. Checking your own systems, tightening access and briefing the board are several more. CyPro's responders and consultants can take any of them on.
Incident response
If you may already be compromised, our responders contain it, preserve the evidence, find out what was touched and get you running again safely.
More on cypro.co.uk
24/7 security monitoring
Our SOC analysts watch your systems day and night for the signs of an attack in progress: unexpected processes, new outbound connections, logins that do not fit.
More on cypro.co.uk
Fractional CISO for the board
A senior security lead who advises the board on what to fix first, what to accept and what to tell customers, and owns the decision record.
More on cypro.co.uk
Penetration testing
A focused test of your systems and the controls around them, so you know they hold before an attacker finds out for you.
More on cypro.co.uk
Hardening and segmentation
Our consultants put the controls in place that limit the damage when something is missed: firewall rules, network isolation and tighter admin access.
More on cypro.co.uk
Vulnerability scanning
Find every internet-facing system you run, including the forgotten ones, and see which are still on a vulnerable build.
See the scanning service
Talk to a specialist
Worried this affects you?
If data has leaked through an AI tool or an exposed AI database, the first hours decide how much is lost and what you must report. An incident response team helps you contain it, work out what was exposed and meet your notification duties.