Latest
IBM publishes its 2025 Cost of a Data Breach research[5]
The study looked at AI security and governance for the first time and reported on the cost of shadow AI and missing access controls.
Overview
AI data breaches: what they are and what to do
An AI data breach exposes data that AI tools take in, store or produce. It is not one incident but a pattern, and the reporting duties under UK GDPR are the same as for any other personal data breach.
Last reviewed against legislation.gov.uk, National Cyber Security Centre and OWASP GenAI Security Project. Sources
At a glance
- 72 hours
- UK GDPR deadline to notify a reportable personal data breach, from becoming aware[1]
- 97%
- of breached organisations with an AI-related incident lacked proper AI access controls[5]
- 63%
- of organisations studied had no AI governance policies[5]
- 1.2%
- of ChatGPT Plus subscribers active in a nine-hour window had payment details exposed[7]
The latest
AI Data Breach: what has happened so far
Newest first. Each entry links to its source.
Wiz reports an open DeepSeek database[6]
Wiz said a publicly accessible ClickHouse database with no authentication held over a million log entries, including chat history and API keys. DeepSeek secured it after disclosure.
NCSC publishes AI and cyber security guidance for leaders[3]
The guidance covers data risks, supply chains, prompt injection and the need for a response plan for serious AI-related incidents.
The full timelineIn brief
Three answers first
What the term covers, whether it applies to you and what to do first.
What counts as an AI data breach
Data can leak at four points: when staff paste it into a public AI tool, when an AI service stores it in logs or chat history, when an LLM repeats it in an output, and when a database or pipeline behind an AI product is left open. OWASP lists personal information, credentials and confidential business data among the things an LLM can disclose [4].
Does it apply to you
If your organisation uses any AI tool, or holds personal data that an AI supplier handles for you, yes. Samsung restricted generative AI tools after engineers uploaded internal source code to ChatGPT [8]. In the IBM study, 63% of the organisations studied had no AI governance policies to manage AI or stop shadow AI [5].
What to do first
Find out which AI tools your staff actually use and what goes into them [3]. Then read what to do this week and how to report an AI data breach.
Real cases
What has happened so far
Public cases show different causes, not one.
A software bug
A bug in an open-source library meant ChatGPT showed some users other users' chat queries and payment details, for about 1.2% of ChatGPT Plus subscribers active in a nine-hour window [7]. See the examples in detail.
An open database
Wiz found a publicly accessible database with no authentication that held over a million log entries, including chat history and API keys [6]. See exposed AI databases.
Staff using public tools
Samsung's memo warned that data typed into AI platforms is stored on external servers and is hard to retrieve or delete [8]. See shadow AI and data leaks.
Start here
One thing to do first
First action
Write down every AI tool in use, who uses it and what data goes in. The NCSC tells leaders to understand their data and supply chains before anything else [3].
What to do
Four things to do first
-
1
Find where AI touches your data
IT, security and data protection leads
-
2
Set a clear rule on what may be entered
Directors, HR and line managers
-
3
Restrict access to AI data stores and logs
IT and cloud teams
-
4
Prepare the breach response
Data protection officer and incident lead
There are six steps in all, with who does each one. See every step
Am I affected?
Answer a few questions
Check if you are affected
Answer three questions about staff use, AI data stores and your incident plan to see where to look first. Nothing you answer leaves this page.
Do staff use AI tools for work without an approved list or a written rule on what data may be entered?
Quick answers
Questions people ask
What is an AI data breach?
It is a compromise that exposes data used by or produced by AI tools, such as prompts, chat logs, outputs or training data. OWASP lists personal information, financial and health records, confidential business data and credentials as data an LLM application can disclose[4]. The legal test is the same as for other incidents: whether personal data was exposed.
Do I have to report an AI data breach to the ICO?
If it is a personal data breach, UK GDPR requires notification not later than 72 hours after becoming aware of it, unless it is unlikely to result in a risk to people's rights and freedoms[1]. If you report late you must give the reasons for the delay[1].
Do I need to record a breach I decide not to report?
Yes. The controller must document any personal data breach, including its facts, effects and the remedial action taken[1]. Keep the reasoning for deciding that it was unlikely to result in a risk.
What is shadow AI?
It is AI use that the organisation has not sanctioned or does not control, such as staff using public chatbots for work. IBM's 2025 study found that high levels of shadow AI added USD 670,000 to the average breach cost[5].
About this site
Primary sources, dated
The CyPro Consulting Ltd team maintains this site. Every version, date and figure comes from 8 named sources, including legislation.gov.uk, National Cyber Security Centre, OWASP GenAI Security Project, and links to the one it came from. Last checked 4 November 2026.
Who we are and how pages are kept currentStart here
The guides, in the order to read them
What to do
Practical steps for UK organisations: find AI use, restrict data, prepare breach response and brief staff, each tied to NCSC, OWASP and UK GDPR sources.
Read the guide
Timeline
The public AI data exposures and the UK guidance that followed, in date order and sourced, from the March 2023 ChatGPT bug to the 2025 IBM research.
Read the guide
Questions
Sourced answers on what counts as an AI data breach, the 72 hour reporting rule, shadow AI, exposed databases and what UK organisations should do.
Read the guideGo deeper
In depth Shadow AI 
In depth Exposed databases 
In depth Prompt leaks 
In depth Reporting 
In depth Examples 
In depth Prevention 
In depth Containment 
In depth Suppliers 
In depth Data protection 
In depth AI use policy How 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.