Response to interpellation 2024/25:658 on countering fraud in the welfare system with the help of AI
Translated from Swedish by AI; the translation may contain errors. The Swedish text is the original.
Summary AI, written in advance
KD argues that the government takes welfare crime seriously by tearing down secrecy barriers and facilitating information exchange 1. KD emphasizes that AI has great potential to improve services and advocates for an AI workshop to combat fraud 1 2. KD agrees on the need for system-wide measures and argues that the government's regulatory framework is effective 3, while KD sees the entrepreneur's work as a source of inspiration 4. S argues that a systems approach is necessary to address the shortcomings 5. S advocates that the Swedish Social Insurance Agency should be allowed to use the police's intelligence register 5 and that state agencies should be given the mandate to eliminate fraud 6. S considers AI solutions to be effective for combating criminal economy 7.
Written by AI in advance and may contain errors. The numbers lead to the speech a statement builds on; check against the text below.
Statsrådet Erik Slottner (KD)
Mr. Speaker! Mikael Damberg has asked me if I have taken any initiatives for analysis to investigate why private individuals succeed in identifying obvious examples of fraud and crime while the state authorities have not done so.
The government takes welfare crime seriously. From the state welfare systems alone, it is estimated that around 15–20 billion kronor are paid out incorrectly every year, of which half is suspected to be due to intentional errors. The government and relevant authorities are working together to stop welfare crime.
The Government has taken extensive measures to counter incorrect payments and criminal exploitation of public funds. The Government is, for example, working on a broad front to tear down secrecy barriers and to facilitate for authorities and other actors to share information with one another. Extensive investigation and legislative work is ongoing. For example, a new law on strengthened information exchange to counter incorrect payments from the welfare systems and crime in the labor market has already been introduced.
It is necessary to have accurate and effective controls to counter welfare crime. Many authorities therefore use data analyses and sampling to identify irregularities. To ensure that the relevant authorities have clear legal support to process personal data for this work, the government has carried out several investigation and legislative projects.
It is also important to close all doors that could allow unserious and criminal actors to have the opportunity to participate in public procurements. In order to more effectively counter criminal and unserious suppliers in public procurements, the government has provided Bolagsverket with funds to be responsible for providing an IT system for coordinated register control of suppliers in public procurements and choice systems. Konkurrensverket has also been tasked to review procuring organizations' possibilities to exclude abnormally low bids that are not seriously intended.
A series of other different measures aimed at increasing the digitalization of public procurement has been presented in the Action Plan against Corruption and Undue Influence 2024–2027. For example, Statens servicecenter was tasked with promoting increased use of digital orders and strengthening the capacity for procurement analysis in the state administration. The assignment was reported on the last of February.
The Government will continue the work of countering incorrect payments and crime in order to strengthen and maintain a strong protection for our common welfare.
AI has great potential to increase the quality of the authorities' services, reduce the amount of administration, improve welfare and strengthen Sweden's competitiveness. The Government has therefore, in the regulatory letters for 2025, tasked 25 authorities to account for their use of AI, including Försäkringskassan and Pensionsmyndigheten. The authorities shall also report on what challenges exist linked to ethics, law and security.
Statistics Sweden reported in March 2024 the report Authorities and AI – A study on opportunities and risks of using AI in the public administration. The report shows that several authorities, including Försäkringskassan, Arbetsförmedlingen and Skatteverket, already use AI to perform risk assessments of cases to detect errors or criminal activity. One example is that all tax returns are checked mechanically, and there AI is an element to find the tax returns where the risk of inaccuracies or crime is greatest. The most high-risk cases then undergo a deeper review.
The Government proposes in the 2025 spring budget an increased appropriation to Försäkringskassan and Skatteverket of 30 million kronor each in accordance with the AI Commission's proposal on the establishment of an AI workshop for public administration. The Government therefore intends to task Skatteverket and Försäkringskassan with investigating the conditions for such a workshop in order to increase the public administration's collective capacity to develop and use AI-based solutions. An AI workshop can create conditions for the public sector to develop and share AI services, which can streamline the agencies' work in combating welfare crime.
Municipal activities must be protected against welfare attacks that threaten the system. The Government has therefore proposed amendments to the Local Government Act which entail clearer internal control and aim to strengthen the protection for certain elected officials in municipalities and regions. The Riksdag has said yes to the legislative changes, which enter into force on 1 July 2025. Furthermore, the referral period has recently expired for the proposals regarding municipal connection to the Swedish Payments Authority's operations, and the proposals are now being prepared within the Government Offices.
Furthermore, the government has tasked Statskontoret with proposing measures to improve municipalities' and regions' control and follow-up of private contractors in order to counter welfare crime, unscrupulous actors, and corruption. Statskontoret's proposals have been sent out for consultation.
Mikael Damberg (S)
Mr. Speaker! Thank you to the Minister for the answer!
The background to this interpellation is that there is a very large amount of criminal economic activity taking place in Sweden. Just a few weeks ago, I believe, the Economic Crime Authority presented a report that estimated the criminal economy in our country to be somewhere between 100 and 150 billion every year. These are enormous sums that are turned over in the criminal economy.
The Economic Crime Authority's analysis is that the problems have become increasingly system-threatening. From involving individual people who cheat or make errors, they have transitioned into a more organized form. It is, of course, important to work on this.
I appreciate that the Minister gave examples of how AI is used today in the public administration and in individual agencies to combat fraud and inaccuracies. It is a development we have seen over quite many years. One has begun to adopt new technology to make the processing both more efficient and more correct, in some sense. But I believe that one must have a bit more of a systems thinking to address the deficiencies.
In my interpellation, I highlighted an example of a person whom I did not name, but I can do so today. Jens Nylander, whom both the Minister and I have met and listened to, is a serial entrepreneur who on his own has requested data on public procurement from authorities and scrutinized the Swedish Tax Agency and annual reports and found a great deal of fraud, inaccuracies, and waste in the public sector.
My questions, which I would like the Minister Slottner to return to, are: Has any analysis been made of what Jens Nylander accomplished? Can authorities do the same in the state administration today? Can they achieve the same results? It is somewhat embarrassing that he, through his own engagement, revealed systematic cheating and trickery, but that the police, Skatteverket, or Bolagsverket have not managed to detect the same thing. I believe there is a development potential for the public sector to work with.
I want to give another example. We presented a proposal a few weeks ago that the Swedish Payments Authority should be given the opportunity to access the Police Authority's intelligence register on gang criminals in Sweden. It is 62,000 people who are included in the police's intelligence register of people who are either gang criminals themselves or who are persons with close links to gang criminality.
Our proposal is that the Swedish Payments Agency should be allowed to use it to be able to pinpoint public payments to the gang criminal environment in order to find more opportunities for fraud and incorrect payments so that it can be choked off.
There must be a systems thinking when considering how to use the information that exists in the public sphere – in registers, in statistics and in other data – to access the criminal economy.
I therefore return to my question: Has any analysis been made of the exciting work that Jens Nylander has accomplished as an individual? I am convinced that both the Minister and I have been somewhat inspired by it. Can the Minister elaborate on what conclusions he draws from this and if there are changes that must be made so that this can also be implemented within and between agencies?
Statsrådet Erik Slottner (KD)
Mr. Speaker! No formalized or public analysis has been conducted. I want to be honest about that. On the other hand, we have, of course, taken part of the information and also met with Jens Nylander to get more information, because it is very exciting.
This testifies to the importance of public and private cooperation. Private entrepreneurs will always be a step ahead of the public sector in the development of digital services. It is a core task for many entrepreneurs to develop just such things. If the public sector is to be successful within digital services and digital development, one must involve private entrepreneurs. That is obvious. This is a concrete example of that. I do not, for my part, think that they need to stand particularly much against each other.
I often speak about how we can make data accessible within public procurement. We lack a system for a national procurement database such as some other Nordic countries have. At the same time, one can obviously obtain procurement data in other ways, as Nylander shows. This is, in any case, an issue that is being discussed.
We are now working on a roadmap for public procurement, also with the aim of promoting more strategic work. It is completely obvious that the competence and the strategic work surrounding public procurement needs to increase throughout Sweden, not least in our municipalities. We have 290 municipalities, and each municipality is responsible for its own procurement.
We see, not least, the importance of digitized procurement processes where one can digitally match an order with an invoice. The County Administrative Board of Västra Götaland has reduced its costs by 15 million just by introducing e-invoices. It is just one concrete example of what digital services mean for preventing and detecting intentional fraud and unintentional incorrect payments.
For some time now, a dozen authorities have had a special mandate to work with welfare, the criminal economy, and welfare crime. I believe it was as a result of our first spring budget that they received both the mandate and extra funds for it. A dozen authorities are, as mentioned, involved in this.
Together with, among others, Minister of Justice Gunnar Strömmer, I have been able to present legislative proposals on how we break confidentiality and remove confidentiality barriers in order to detect welfare crime. It will now be much freer than today to be able to share information between different levels of authority and between different state agencies. It is a very important measure.
Then we have the proposal from the AI Commission regarding an AI workshop. The idea is to develop and test various AI services that can be used jointly in the public sector. This could, for example, involve developing AI services to detect welfare crime and fraud. These could then be used broadly in the public sector. That would be very good.
As I said in the interpellation response, we already see today that, among others, Försäkringskassan and Skatteverket use AI to detect incorrect payments and fraud or outright criminal activity. These tools could certainly be used much more broadly in the public sector, beyond the two mentioned agencies.
Mikael Damberg (S)
Mr. Speaker! I would like to once again thank the Minister for Civil Affairs for the answer but also for his reflections on what is being done today.
Let me speak a bit from my own experience. How many times have we not made decisions to break confidentiality between or within agencies? And two or three years later, it turned out that it was not enough to achieve change. We give assignments to agencies to work with AI development or to combat fraud, even though we know that a large part of the solution does not really exist within the agencies but between agencies. That is how Jens Nylander actually saw the fraud.
One thing is a bit strange. I do not think the Swedish people know that Bolagsverket's annual reports are not audited by anyone. Skatteverket, for example, does not check if their figures on taxes paid from the companies match what is in Bolagsverket's annual report. This is not entirely optimal.
There is a lot of information at the state agencies that is not compared with the information at others. It is actually this that this entrepreneur has done. He has requested a lot of documents, run AI services, and noticed that the data did not match. It is such systematic work that needs to be done within state agencies. It really needs to be done.
The same applies to the latest proposal, which I would like the Minister to comment on. We proposed that a payment authority should be established and initiated an investigation, which you have since continued to develop, which is good. I believe that the Payment Authority is a way to get control over payments going to people all over Sweden so that we can avoid fraud and inaccuracies.
But if the Payment Authority could use the police's intelligence regarding gang criminals, it could very easily conduct checks to see if payments from different authorities have gone to individuals whom the police suspect are involved in gang criminality. Then, one can direct the control measures towards exactly these individuals. By doing so, we would make it incredibly much harder for gang criminals to exploit the public system.
I believe it is this kind of authority-transcending and more strategic use of information that needs to come about for us to address the problems.
I do not primarily want a political discussion about this, because it is a common problem for governments over time. We have all wanted to tackle fraud and inaccuracies but perhaps have not received enough return on the work.
Now we will have the AI solutions that the respective agencies use. We must find more models to use it broadly between agencies. I believe it would be a very effective way to access criminal economy. If we do that, we can save money that otherwise goes to incorrect payments. It is about many billions. We could access the criminal economy, which Bolagsverket warns about.
Not least as a former Minister for Finance, I have a very great interest in the black economy becoming white. Then we would receive tax revenues that we can do something sensible with instead.
I hope that the Minister for Civil Affairs sees this as an input into the work. He is welcome to reflect on the more concrete questions I am asking regarding how authorities can work together to access the criminal economy.
Statsrådet Erik Slottner (KD)
Mr. Speaker! I completely agree with Mikael Damberg when he says that we need to work with system-wide measures. It is a development we have seen over a number of years now, with a criminal economy that is growing and which is going deeper into more and broader welfare sectors. As we have said before, public activities require system-wide work.
It is also the background to many of the decisions that the government has made. Even though the previous governments have removed some secrecy barriers, it is not at all about anything other than that what we have now proposed in three steps is very far-reaching and much more comprehensive than the previous legislative changes that have occurred.
I truly believe that the confidentiality-breaking regulatory framework we last were able to provide information on regarding these authorities, which also includes municipalities and regions, is very far-reaching and will have an effect. Regarding the dozens of authorities I spoke about earlier, this is also another way to gain a more unified grip.
I also want to mention that the roadmap for public procurement, which we are now working on, is also a way to bring in Kommun- och regionsverige.
A concern that I encounter quite often as a minister responsible for both the public administration and the relations with municipalities and regions is that we do not govern. We have in Sweden an extensive local self-government – that is what we agree to have. But this sometimes creates challenges because we do not govern the municipalities with regulatory letters. We must therefore work in a different way and, not least, have a close cooperation with SKR in these matters.
I also want to mention the digitalization strategy, which the government will be able to make decisions on within a fairly short time. It also involves a holistic approach. The operations of the municipalities and regions will be covered by the digitalization strategy, and SKR has been involved in developing the strategy, which I very much welcome.
The AI workshop is an example of how we can develop common solutions between agencies, which can also be used at different levels in the public administration. It is also a system-wide way of working.
I mention these examples to show that I agree with Mikael Damberg regarding what he says about the fact that we need to work more system-wide. We have begun such a journey, even though we will need to do even more.
I also want to mention the Supplier Control Investigation, whose proposals we have now incorporated into the budget. Bolagsverket is working on a supplier control so that before an agreement is concluded, one can detect errors and deficiencies through a collective supplier register that Bolagsverket will be responsible for.
When it comes to the specific input regarding the fact that the Swedish Payments Agency could be allowed to receive the police's intelligence information, I think it sounds like an interesting proposal. But I still have to pass when it comes to the answer, because it is Finance Minister Elisabeth Svantesson who is responsible for the Swedish Payments Agency. The Swedish Police Authority is Justice Minister Gunnar Strömmers responsibility. I would perhaps like to let Strömmer and Svantesson look more into that proposal. But I think, without a doubt, as I stand here and listen to the proposal, that it is an interesting input that we can certainly look into further in the future.
Mikael Damberg (S)
Mr. Speaker! Thank you, Minister, for the interest in the proposal! I have initiated a dialogue with the Minister for Finance regarding this, but I have not received such clear answers so far. But when new proposals are raised, one often has to reflect a bit and turn them over and over.
I believe, however, that there is an opportunity here. With the new Payment Authority, which hopefully can soon also include the municipalities' payments, which I believe we are both involved in, there is a truly powerful tool to actually access the criminal economy in a smart way. But then they must work in a different way than they do today in order to be able to direct the search efforts towards precisely the criminal economy and gang crime in Sweden. I hope that you investigate this and look further into it.
My question was about the analysis of how this entrepreneur had worked. I appreciate many of these initiatives; I believe many of them are good. I believe that the confidentiality-breaking provisions that are now coming, which we began investigating and you have taken further, will be significant for the authorities' work. But I am still somewhat skeptical. Can an authority, say the Swedish Police Authority, do what Jens Nylander, a private individual in Sweden, has done to abolish cheating, fraud, and crime in our country? I do not believe they can do it today, and I think it is strange. If it is not to be the Swedish Police Authority, let the Swedish Tax Agency or some other authority be given the mandate and opportunity to do what Jens Nylander has done. State authorities should be able to do that.
I appreciate that entrepreneurs lead the way and go first. I have no problem with that. But in a matter that is so central to the state, which concerns credibility and also the public economy and the fight against crime, the Police Authority should have the muscles and opportunities that this entrepreneur has had. Does the Minister of Justice think that the Police Authority could do what Jens Nylander has already done?
Statsrådet Erik Slottner (KD)
Mr. Speaker! I will nevertheless refrain from guessing in the interpellation debate. Now, many different authorities are also involved here, including the Swedish Police Authority, which does not fall under my area of responsibility. Therefore, I am being a bit extra cautious about reflecting too far-reaching, which I can sometimes dare to do when it concerns any of the 35 or so authorities that I have on my desk.
I welcome the debate. I don't actually perceive any great polemic here, rather we both want the same thing in this issue. In this debate, I have accounted for the measures that we have taken so far, which are also system-wide and interdepartmental.
I am, however, completely convinced of one thing: We will need to do much more. I believe that Mikael Damberg and I agree on that. Reality moves fast, and the criminal gangs and groups have a tendency to run very fast, many times faster than politics and legislation, so that we cannot keep up. It is a concern.
We also have long preparation times. Damberg mentioned that it takes time to reflect on new proposals. Then there is also joint preparation and other things that need to be done, and many legislative proposals need further investigation. Such things take a lot of time, which is naturally a concern in the administrative model that Sweden has and has had for several hundred years.
But with that said, we have developed the measures I have accounted for here very quickly. There has been an enormous effort, not least in Gunnar Strömmers area, to secure the necessary legislative changes to be able to counter welfare crime. As minister responsible for public administration but also for digitalization and public procurement, I have a very large responsibility within those areas to truly use the tools as well as we can. I believe I have accounted for some of this in my response.
I still choose to see entrepreneur Jens Nylander's work as a great source of inspiration regarding how the public sector could also work.
Source: The Swedish Parliament. The speeches come from the open data of the Riksdag, translated into English by AI, which may contain errors.