How does the IND use artificial intelligence (AI)?

AI is developing super-fast, as is the public debate surrounding it. How does the government use AI? And the IND? Do AI applications play a role in decisions on residence permits? IND work concerns the lives of people. This is a major responsibility and requires openness about how the IND uses this technology in its work. Because of that, the IND has therefore developed a vision for the use of AI.

No high-risk AI systems at the IND yet 

In Europe, rules govern the use of AI by public authorities and other organisations. These rules are laid down in the EU AI Act. The greater the impact on people, the stricter the requirements. Decisions on residence and asylum directly affect people. For this reason, AI applications that play a role in such decisions could fall into the ‘high-risk AI’ category.

At present, the IND uses AI only for support tasks. Examples include VisualDOC, which helps to recognise and classify documents such as birth and marriage certificates, and the Sofia chatbot, which helps visitors to the IND website find the right information. These AI tools are not classified as high-risk AI systems. However, the IND is researching applications that potentially could become high-risk. One example is the CELIA project, which focuses on innovation and optimisation in language and dialect recognition for asylum seekers, combined with human expertise. This system must first clearly demonstrate that it complies with all the strict European rules.

The IND also uses algorithms that are not AI: these systems operate based on fixed, pre-programmed rules and do not learn or adapt by themselves. This technology helps our staff carry out administrative and repetitive tasks more quickly. Our staff always decide on an application. 

Opportunities in practice

The IND is exploring new ways in which AI can support its work. For example, it tests if speech technology can be used to transcribe interviews. This could save staff time. The IND is also developing applications that can summarise documents and case files more quickly, so staff can identify the core issues of a case sooner. This makes the work more efficient and is also in the interest of the applicant. 

For example, since the summer of 2026, the IND has been testing the AI assistants Robin and ChatIND internally on a limited scale. Robin is being tested by around 100 staff members and ChatIND by around 130. Both have been classified as ‘low risk’ by the Algorithms Assessment Committee.

The AI assistant Robin is a digital assistant that helps IND staff in their day-to-day work. IND staff can ask Robin questions, and Robin helps them to draft, structure and improve texts more quickly. Robin can also search and summarise documents extremely quickly, enabling staff to find the information they need. The central principle is that Robin may only be used to process non-confidential or publicly available information. Robin may not be used for decision-making.

ChatIND is a similar system that operates within its own secure (on-premise) environment of the IND. Confidential documents can be processed using the system, such as a summary of a legal case file, for example.

The evaluation of ChatIND and Robin still needs to take place. Once they are put into use in the day-to-day practice of the IND, the IND will report this to the national Algorithm Register.
 

Seven principles for the use of AI

The IND has adopted seven principles to guide the role of AI within the organisation. Here is a summary:

1. We ensure that our staff are AI-literate. This means that IND staff who work with AI receive training and learn what an AI system does, how to assess its outputs and what its limitations are. 
    
2. We use AI to support our ambitions, values and processes. This means that we start from our work processes and consider where AI can improve them. AI is a means, never an end in itself. 
    
3. We assess AI applications in advance to ensure that they are ethically and legally sound. For higher-risk applications, we carry out a Human Rights and Algorithms Impact Assessment. This is a mandatory assessment through which the government checks in advance if an algorithm or AI system operates safely, fairly and in accordance with human rights.
    
4. We assign responsibilities and monitor them throughout the life cycle of AI applications. This means that someone is always ultimately responsible for ensuring that an AI application works properly, from beginning to end.
    
5. We reduce the risks and consequences of bias (e.g. prejudice or partiality) and incorrect use. Data is never neutral, and algorithms can unintentionally treat certain groups differently. We already take this into account when designing and developing AI applications and continue to do so by monitoring and raising staff awareness of this risk. 
    
6. A person is always responsible for a government decision. An AI application or algorithm may play a role in preparing a decision, but a member of staff makes the decision and can explain how it was reached. This human intervention must be a meaningful part of the decision-making process.
    
7. We communicate openly and transparently about the use of algorithms and AI. For example, the IND publishes information about algorithms and AI applications in the government’s Algorithm Register, even when it is not required to do so.

Cooperation

The IND does not tackle the use of AI in isolation. Within the Ministry of Justice and Security (JenV), the IND works with the AI Directorate and other executive agencies on AI-related topics such as AI literacy and transparency. The IND also works with the JenV’s CDO Office on joint guidelines and frameworks to meet the requirements of the EU AI Act. Cooperation with knowledge partners such as the Netherlands Organisation for Applied Scientific Research (TNO), the Netherlands Forensic Institute (NFI) and universities also plays an important role in this regard.