Home / Glossary / DLP (data loss prevention)
Glossary
What is DLP (data loss prevention)?
DLP (data loss prevention) is a set of tools and rules that find sensitive data and stop it leaving the company without permission. It covers data at rest, in motion and in use, which means data in storage, on the network and on the device. DLP was built for email, files and web uploads, and AI agents now move data in other ways.
By the AIBound team · Updated 17 November 2026
Key takeaways
- DLP finds sensitive data and applies a rule when someone tries to send, copy or upload it.
- It covers data at rest, in motion and in use.
- DLP was built for email, files and web uploads, so data that moves through an AI agent needs a check inside the agent too.
Why DLP matters
Every company holds data it must keep in: customer records, payment details, health information and source code. Privacy laws and customer contracts expect the company to know where that data is and who can send it out. Most leaks are mistakes, such as an email to the wrong person or a file saved to a personal account. DLP catches these at the moment they happen.
AI adds new paths. When an employee pastes a customer list into an AI app, that looks like a web upload, and DLP can read it. When a coding agent reads a file of secrets and posts it to an outside host, that is an action taken by the agent. It does not look like a file upload, so it calls for a rule that runs inside the agent.
How DLP works
DLP needs to know what counts as sensitive before it can protect it. The steps are the same whether it runs on the network, in the cloud or on the device.
- 1 The company defines what is sensitive, such as card numbers, health records, source code or files with a label.
- 2 DLP scans storage, email, network traffic and devices to find that data.
- 3 When data moves, DLP compares the content and the context against the rules.
- 4 DLP allows the action, warns the employee, encrypts the data or blocks it, and logs what happened.
DLP vs AI DLP
DLP is the long-standing category. AI DLP is the newer name for the same goal applied to AI: keeping sensitive data out of prompts, tool output and the actions of AI agents. A company that owns DLP keeps it and adds the AI part.
| Question | DLP | AI DLP |
|---|---|---|
| What does it watch? | Email, files, web uploads and removable drives | Prompts, tool output and the actions of AI agents |
| How does the data leave? | A person sends, copies or uploads it | An agent reads it and sends it on as part of a task |
| Where does the control sit? | On the network, in the cloud and on the device | Close to the AI, in the prompt or inside the agent |
Sources
- NIST glossary: data loss prevention · csrc.nist.gov
Sources checked 17 November 2026
Examples
DLP examples
FAQ
Frequently asked questions
What is DLP (data loss prevention)?
DLP is a set of tools and rules that find sensitive data and stop it leaving the company without permission. It checks data in storage, on the network and on the device. When someone tries to send, copy or upload protected data, DLP can warn them, block the action and log it.
What are the three types of DLP?
Network DLP reads data as it crosses the network, such as email and web traffic. Endpoint DLP runs on laptops and desktops and watches copying, printing and uploads. Cloud DLP checks the data stored and shared in cloud apps. Many companies use all three, managed from one set of rules.
What is the difference between data loss and data leakage?
People often use the two terms for the same thing. Where they are split, data loss means the company no longer has the data, for example after a deletion or a ransomware attack. Data leakage means a copy reached someone who should not have it. DLP tools mostly deal with leakage.
Does DLP stop data leaking through AI tools?
DLP can read a paste or an upload to an AI website, the same as any web upload. An AI agent that reads a secret and acts on it in one step is a different path. AIBound enforces policy inside the agent, so a credential read is denied in real time.