Police GPT

Police

PolitieGPT is an in-house chatbot developed for police officers. The chatbot uses generative AI. Officers can use PolitieGPT to ask questions about internal documents, policies and guidelines. PolitieGPT also helps with rewriting texts, generating text suggestions and summarising information.

Last change on 14th of June 2026, at 11:02 (CET) | Publication Standard 1.0

• Publication category: Other algorithms
• Impact assessment: DPIA, Ethics quickscan, Review of the AI Regulation
• Status: In use

General information

Theme

Public Order and Safety

Begin date

2026-05

Contact information

https://www.politie.nl/

Responsible use

Goal and impact

PolitieGPT helps staff to do their work more quickly and easily. Staff can query and edit documents. PolitieGPT can also assist with writing and rewriting texts, generating text suggestions, summarising documents and translating information.

PolitieGPT is intended to support internal operational management and police duties. The chatbot is not intended for processing police data covered by the WPG. The focus is entirely on internal processes within the police force.

Considerations

Generative AI is being used more and more frequently, and many staff members are benefiting from it. As an employer, the police aim to provide clear guidelines for the responsible and authorised use of this type of technology. That is why they have opted for a secure internal chatbot.

With PolitieGPT, the police can better protect data security and monitor usage. At the same time, this technology offers opportunities to support and improve their work. The police wish to make use of these opportunities in a responsible manner.

Human intervention

The officer asks the system questions or uploads a document. The system provides answers based on that document or on internal police policy. If an answer is based on policy, the system also shows where that information comes from.

The system does not make independent decisions. The officer remains responsible at all times for the use and outcomes of PolitieGPT.

Risk management

To prevent misuse, clear limits have been set for PolitieGPT. For example, the system is not permitted to provide medical information to staff.

Staff are only granted access to PolitieGPT after they have completed training on the use of AI. All queries and requests made to the system are recorded for subsequent review.

Security for PolitieGPT is being further strengthened. This is being done in accordance with the BIO standard, the national standard for information security within the public sector. In addition, clear agreements on the use of data are being drawn up. These will specify who is responsible for what and how inputs, outputs and knowledge sources are to be handled.

Legal basis

The European AI Regulation forms the basis of police policy regarding the responsible development and use of AI and algorithms.

Links to legal bases

AI Regulation: https://eur-lex.europa.eu/legal-content/NL/TXT/PDF/?uri=OJ:L_202401689

Impact assessment

• Data Protection Impact Assessment (DPIA)
• Ethics quickscan
• Review of the AI Regulation

Operations

Data

The police use open-source models in accordance with government guidelines. The model has not been further trained using additional police data.

Clear agreements have been established regarding the use of data and the system’s outputs. These agreements describe how data is handled and who is responsible for what.

Technical design

PolitieGPT runs on the police force’s own infrastructure. The system uses an open-source language model that operates within the police environment. This ensures that sensitive and internal data always remains within the organisation’s own network.

The user works via a chat window, where they can ask a question or upload a file. The system sends the question to the language model. The model processes the text and generates a response, which is then displayed to the user in the chat window.

The system does not learn autonomously and cannot retrieve information from the internet. For questions regarding police-specific information, the system uses internal knowledge sources (RAG pipelines). This enables it to provide targeted and appropriate answers.