AI service to strengthen investigative journalism

Photo: Tim Mossholder, Unsplash.
Can AI make it easier for journalists to scrutinize public documents from local authorities and government bodies? A new research project at Jönköping University is developing an editorial service designed to collect, structure and assess the newsworthiness of public documents. The aim is to strengthen journalism and democratic accountability.
The project is led by Malin Picha Edwardsson, Assistant Professor of Media and Communication Studies at the School of Education and Communication at Jönköping University. She believes that the investigative role of local journalism has become increasingly difficult to maintain.
“Democratic processes involve the media scrutinizing local politicians, authorities and regions to understand decision-making processes and help influence democratic decisions. But with reduced resources, local media have limited opportunities to carry out this monitoring in a structured way. Our new AI tool can therefore provide greater opportunities for effective scrutiny,” she says.
Prototype to be tested in newsrooms
The development of large language models opens up new possibilities for reading, sorting and assessing documents such as minutes, notices, appendices and briefing materials.
The research project begins with interviews with local government reporters and other staff at five selected pilot newsrooms. The aim is to map out how local government coverage is currently conducted and what needs exist. An initial prototype will then be developed and tested in the newsrooms.
The prototype is designed to collect and structure public documents, carry out a news assessment to identify potentially interesting cases, and summarise the content. The newsrooms will then decide which cases they wish to investigate further.
"The idea of automating the monitoring of local decision-making has been around for many years. Rapid advances in AI have given us more powerful tools than ever before for reading and analysing large amounts of unstructured data. At the same time, we can build new tools much more efficiently than before with AI support," says Jens Finnäs, CEO of the news agency Newsworthy, which is one of the project’s partners.
Great potential for making decisions more transparent
The prototype will be tested in collaboration with the pilot editorial teams, focusing on aspects such as accuracy, quality and usability. A key part of the feasibility study is also to analyse risks, such as how AI handles sources, uncertainties and the risk of inaccurate summaries.
Malin Picha Edwardsson believes that, in the long term, the project could strengthen local journalism and democratic transparency in municipalities and regions where coverage is currently limited. The plan is for the preliminary study to lay the foundations for a larger research and development project on AI in journalistic practice.
“If we can use the technology responsibly, then there is a real opportunity to highlight decisions that would otherwise risk slipping under the radar. We are investigating how AI can be used as an editorial aid, but it should of course not replace journalistic judgement,” says Malin Picha Edwardsson.
Project partners
The technical aspects of the project are being carried out by the companies Newsworthy and Kvantab. Newsworthy is a news agency specialising in local data journalism, with around 80 subscribing media companies in Sweden. Kvantab is an AI company that collects public documents from Swedish institutions and structures the content into short summaries.
The project is also supported by the national collaboration platform Media & Democracy at Lindholmen Science Park.
Funding body: The Hamrin Foundation