The democratic challenge

All Danes have a right to public access for all official cases, as safeguarded by the Freedom of Information Act, Section 7. Similarly, any party involved in a case can request to see all related documents, as stipulated by the Data Protection Act, Section 9.  

However, this right to information isn’t absolute. 

Certain data, such as private details of individuals or trade secrets, may need to be anonymised. 

At present, most anonymisation tasks are carried out manually, making the process both time-consuming and error-prone. Applicants sometimes receive sensitive data that should have been redacted, often receiving this data late due to delays in the anonymisation process. These inefficiencies ultimately erode public trust in governmental bodies. 

Surveys conducted by several local councils suggest that the use of AI could cut the time spent on each Freedom of Information request by up to 50%. 

The solution

Netcompany offers a digital solution that employs AI to replace the laborious manual method with automated anonymisation, overseen by human supervision. The tool is expected to markedly enhance quality, as well as improve the management of tasks and processes. 

This solution can be easily adopted by any authority responsible for anonymisation tasks. In practical terms, the tool combines AI with human oversight. The AI model automatically identifies sensitive elements within sentences and substitutes them with pseudonyms such as ‘party,’ ‘case officer,’ ‘guardian,’ and ‘address’. 

Each document is then fine-tuned and quality-checked by a case officer, who approves, modifies, or adds pseudonyms, whilst also making exceptions in the text where legally permitted. These adjustments help to continually improve the AI model. 

All anonymisation tasks are displayed and updated within an easily navigable case management module. 

A similar tool is already in use by the Judiciary Administration, achieving a 91% accuracy rate in identifying sensitive personal data that must be redacted before documents can be made public.

• Saves case officers valuable time by accurately and automatically identifying sensitive elements within documents
• Standardises the rules and processes governing Freedom of Information requests
• Ensures consistent quality in handling Freedom of Information requests and offers the potential for system-supported quality control
• Rationalises the naming conventions used across documents, thereby enhancing readability
• The AI model facilitates ongoing improvements in identifying sensitive personal data
• Boosts public trust in governmental bodies by reducing the likelihood of errors
• In the long term, the system is likely to yield financial savings