Your contact for new licenses
It’s summit time again – this time not in Brussels, but in Frankfurt. And it is not heads of state and government who are meeting, but experts from politics, business, science and civil society for the Digital Summit 2024. The EU Commission is also present in Frankfurt.
This morning, Prabhat Agarwal, Head of the Digital Services and Platforms Unit at DG Connect, will be among those discussing the results of two years of the Digital Services Act. Hate speech and disinformation pose a challenge for politicians, especially in the super-election year 2024. Just like ahead of next year’s general election. Are the instruments of the DSA sufficient to develop the necessary clout against the major platforms? This will be one of the questions to be discussed.
However, the dominant topic at the digital summit, running until Tuesday, will be artificial intelligence: The topics range from “AI made in Europe – sovereign, trustworthy, secure” to “AI in daily business.” And, of course, the question of the correct regulation of AI will also be on the agenda. According to the experts from the KIRA Center think tank in Berlin, Germany must finally pool its expertise in this area.
Meanwhile, the EU is moving forward with the implementation of the AI Act. This provides for the establishment of a scientific committee of independent experts. The panel is to advise and support the EU AI Office and the national market surveillance authorities in the implementation and enforcement of the AI Act. To this end, the Commission has now submitted a draft implementing regulation and is now asking for comments by November 15. Perhaps you also have some ideas.
Have a good start to the week,
Anyone reading the Draghi report on telecommunications will find many of the demands that the major network operators have been making for years. The report sees an urgent need for consolidation and the promotion of European champions. This is necessary in order to achieve higher investment rates in high-performance networks. Less regulation at the national level and harmonization of spectrum licenses in mobile communications are also necessary.
In fact, the European market is fragmented, while large providers have emerged in other regions. In the EU (450 million inhabitants), there are three to four mobile network operators (MNOs) per member state. In the USA (335 million), there are only three large MNOs nationwide, as there are in China (1.4 billion). French President Emmanuel Macron summed it up in a simple formula: Bigger players can invest more, which means more innovation. However, experts doubt that this equation is correct.
The EU Commission’s Directorate-General for Competition presented its own report in the summer – and came to different conclusions: According to the report, mergers of mobile providers generally led to higher prices for consumers. Positive effects on investments in networks or the introduction of innovative technologies, on the other hand, were not clearly demonstrated. The average revenue per user (ARPU) in Europe is far below that of other regions: In 2022, Europeans spent an average of €15, compared to €42.50 in the USA, €26.50 in South Korea and €25.90 in Japan.
The analysis also shows that European countries with fewer network operators tend to have higher ARPUs. At the same time, there is a negative correlation between market concentration and investments. A larger number of network operators therefore appears to be associated with higher investments and potentially lower prices, while concentration leads to an increase in prices without significant benefits for investments or service quality.
This is in line with the observations of Tomaso Duso, Head of the Department of Enterprises and Markets at the German Institute for Economic Research (DIW Berlin) and Chairman of the Monopolies Commission. “The Draghi Report is very scientifically sound in its analysis in many areas. But the observation that market concentration leads to higher investments is not empirically proven for mobile communications markets,” says Duso. “In fact, we were able to show that joint investments at market level are higher in markets with more providers than in markets with fewer providers.”
Providers, on the other hand, argue with economies of scale. Markus Haas, CEO of o2 Telefónica, put it in simple terms at an event in Berlin: E-Plus and o2, the two smaller providers on the German market, used to invest in two different networks. Since the merger, they have become one competitive network. Instead of investing €500 million each in two networks, o2 Telefónica is now investing €1 billion in a single network.
However, in other national markets, the Commission’s competition watchdogs have not permitted such mergers. In Germany, too, the condition was that there had to be a new fourth provider (which, however, did not get off the ground). The result is an enormous fragmentation of the market.
“As a consequence, we see that the return on investment is too low for the large investments that will now be necessary over the years in various mobile and fiber optic infrastructures,” says Roman Friedrich, Partner at the consulting firm BCG. A BCG analysis shows: “Most telecommunications companies in Europe do not earn their cost of capital.”
However, the reason is complex. It could be that the price level is too low or that there really is a lack of innovation. “But initially the cost of capital is high because we had high interest rates,” says Friedrich. However, Macron’s equation goes a little further. The conclusion that innovative strength increases with size cannot necessarily be drawn from this. “Sufficient earning power may be a necessary prerequisite. But that is not enough to be innovative. It takes more than that.”
His evidence: even in countries with higher profitability, the real innovations do not come from the telecommunications companies. “The innovations come from the digital companies – these are not the existing or past champions,” he says. Duso is convinced: “The innovations come from network technology.” And in this area, Europe already has two European champions in Nokia and Ericsson.
“Of course, it is difficult to get an innovation to scale in a small market,” says Friedrich. However, larger telecommunications companies that can scale across national borders are not necessarily more innovative – not even in the USA. “Innovation culture is a corporate culture, a management characteristic and therefore independent of size.”
For Duso, cross-border mergers could indeed bring advantages. However, this would not require a revision of merger control. The existing regulations do not stand in the way of this. However, he considers it downright dangerous to facilitate non-cross-border mergers and to define the telecommunications market EU-wide for this purpose. The markets have so far been structured nationally or regionally: “The demand side can only access domestic offers locally,” explains Duso. An EU-wide market definition would therefore facilitate mergers, which would lead to higher prices and lower investments.
The competition authorities have fewer reservations about cross-border mergers per se. Some large companies such as Deutsche Telekom or Vodafone have already expanded and become important players in several member states. However, the experiences with this are mixed.
The complexity of such a transnational investment is great, says consultant Friedrich. “You simply need a lot more time and intelligence if you want to implement a scaling model in Europe than in the USA.” There are huge differences on both the capital market and regulatory side.
Friedrich can therefore understand the demand for a single internal market, as well as an EU-wide harmonization of frequency allocation, which companies would like to see. Duso also suggests promoting initiatives to create a more integrated market, for example by harmonizing frequency allocation procedures.
A market structure such as that in the USA would only be desirable if a genuine European market also existed. Before that, a simple comparison of the quality of mobile networks here with those in the USA, for example, is not possible across the board, says Duso. “There are a lot of differences in Europe. There are countries with fantastic networks, not just the Netherlands, but also Sweden and Switzerland. And there are countries with poorer networks. Unfortunately, Germany is one of these countries.” Despite Germany being the largest mobile communications market in Europe.
Mr Fooy, the Swedish Ministry of Justice created the Psychological Defense Agency (PDA) in 2022 to combat foreign false information. What was the reason for setting up this new agency?
In 2014, right before Russia’s annexation of Crimea, we noticed an increase in new forms of disinformation and propaganda. This made restarting a psychological defense capability more urgent. In 2019, the government announced the creation of a new agency for psychological defense. We can think of it as part of Sweden’s broader defense against hybrid threats.
The Swedish government pointed out that Russia, China, and Iran are countries known for disinformation campaigns targeting Sweden. What makes their approaches unique or particularly concerning?
I would say that broadly speaking, Russia has the most experience in reinforcing polarization and segmentation within society. China, of course, has the most significant resources to target vulnerabilities. And Iran is somewhat of a late bloomer. We recently had a very serious campaign against our Child protective services. Various groups in the Middle East, which are not directly state-affiliated, claim that the Swedish protective services were abducting Muslim children and trying to turn them into Christians or turn them over to be raised by homosexual parents. We and others handled this very well, but it could have led to dire consequences.
What does your authority’s work look like when confronted with such “foreign malign information influence” attacks, as you call them?
First of all: We are not an intelligence agency. We work entirely with open sources and provide reports on an operational basis to our government. Much of our work is aimed at building confidence within the Swedish population. A lot of our efforts focus on building public trust, educating people, and collaborating with social services, religious organizations, and other key institutions. However, we don’t aim to counter every piece of disinformation from Moscow, Beijing, or Tehran. That would be a superhuman effort and sometimes even counterproductive.
What challenges do you face especially when it comes to China?
We have to keep in mind that in authoritarian regimes, most information influence activities target the local population at home. That being said, however, China has vast manpower to become active on a global scale. The Chinese state media has an incredible reach. They are published in dozens of languages and are often incredibly professionally produced. I’ve seen cases where reputable outlets like the BBC unknowingly cite Chinese state media sources. Their influence is growing, especially in the Global South. For example, a Chinese publisher bought publishing rights to most of the in-flight magazines in Africa.
How does that affect Western democracies like Sweden?
As mentioned, China’s influence activities increasingly target topics beyond its borders. After the Israel-Hamas war broke out, we noticed anti-Semitic narratives in Chinese state media, for example. It’s peculiar, as anti-Semitism isn’t a common sentiment among Chinese citizens. However, these media reports can be read by Arabic speakers across the Middle East, North Africa, and also Europe, mainly without the readers realizing the source is Chinese. The same thing happens with Russian sources, and many people consuming this content might not question its origin. We want to change that.
Sometimes, it feels like China and Russia are coordinating their disinformation campaigns.
No, I wouldn’t say they coordinate. They do have agreements, like journalist exchange programs, and they amplify each other’s messaging at times. For example, narratives in Russian media may appear in Chinese outlets and vice versa. But calling it “coordination” would be an overstatement. Both are united by the perception of shared external enemies, but their operations are not fully collaborative in the Western sense.
How do you balance protecting Sweden from disinformation while maintaining freedom of speech, especially when Swedish citizens might inadvertently spread false information?
Freedom of speech is critical, and we don’t engage in censorship or fact-checking. That’s for the media and fact-checking organizations to handle. However, we focus on media literacy, ensuring that the public can critically evaluate what they’re reading and understand the intent behind certain narratives. We work with educational institutions to include media literacy in school curricula and have trained over 20,000 individuals on how to identify and counter foreign disinformation.
The PDA also prepares Sweden for the possibility of war. How would the agency support the government during a national security crisis?
While I can’t discuss wartime planning in detail, I can confirm that our mandate includes preparing for all scenarios, from crises to full-scale war. We’re part of Sweden’s total defense strategy.
I’m sure there’s much to learn from the situation in Ukraine and the Russian disinformation there.
Yes, absolutely. The situation in Ukraine is of paramount importance here, as well, because it is of great concern to our government and the Swedish population. And when it comes to figuring out how Chinese information flows works, we also look at Taiwan. But every democracy has a piece of the puzzle to help us understand how these state actors operate.
What challenges do you foresee in the future, for example, due to the advent of new technologies?
AI and deepfakes will be significant challenges, especially in terms of how much information can be flooded into the system. But we do have some advantages – Chinese disinformation, for example, is often constrained by the CCP’s rigid doctrines. Russia, on the other hand, was more imaginative during Crimea’s annexation than in its current full-scale invasion of Ukraine. The quality of disinformation varies depending on the topic and who’s behind it.
So, does that mean the quality differs significantly, even if it comes from the same state actor?
Most state actors do not have one agency dealing with information influence activities. There might be certain think tanks and security services that are part of the armed forces. There’s also a market for disinformation campaigns in China where regular nonaffiliated nationalists all of a sudden get very upset about some trope or narrative in the media space, and they do not necessarily follow the will of the country. So my greatest concern here is that somebody or a number of people in one of the authoritarian states starts being really imaginative and gets lots of resources from their governments. Such an entity could create activities that target vulnerabilities in our society that we aren’t even aware of yet.
Frederick Fooy has been working against information influence activities from various antagonistic actors for more than a decade. He has previous experience from the Swedish Armed Forces dating back to 1989, including deployments to the Balkans. Frederick has a B.A. in Byzantinology from New York University and an M.A. in Urban Policy from Brooklyn College.
The progress made in recent years has turned the former niche topic of artificial intelligence (AI) into a central aspect of economic and security policy. Many countries are responding to this with considerable determination and speed: for example, by quickly and unbureaucratically setting up AI security authorities, recruiting top technical talent and through close exchange with experts from industry and science in specially convened committees. There is currently far too little of this determination in Germany – although we urgently need it now that, among other things, a future-proof implementation of the European AI Act is pending.
The most computationally intensive general-purpose AI (GPAI) models in particular, such as GPT-4, will bring drastic changes to our society: Many experts predict potentially far-reaching effects on productivity or the education system, but also risks such as misuse. This can only be handled with care by those who develop a deep understanding of the technology and its political implications.
This understanding must extend far beyond the Federal Network Agency or a possible new digital ministry to all affected topics: Every specialist policy area must be able to assess opportunities and risks instead of being permanently reliant on outside expertise.
To achieve this, we need to draw on the existing pool of AI experts, many of whom have a German degree. Other countries are already doing this: the UK’s Frontier AI Taskforce (now the AI Safety Institute) managed to recruit top AI researchers with a total of 150 years of scientific experience in just a few months.
Competition for talent is fierce; remuneration and typical career models of the German administration can hardly keep up. Here too, other countries are more flexible: for example, research scientists at the British AI Safety Institute can earn the equivalent of more than €170,000 a year. In addition, fast tracks are being created and formal hurdles removed.
In the pioneering AI paradigm of recent years, deep learning-based general-purpose AI, Germany is only in second divison globally. Virtually all leading GPAI models are currently being developed in the USA, the UK and China. To ensure that we do not miss out on the next developments, Germany now needs bundled expertise.
The German government should therefore convene a committee of leading international experts and provide it with the necessary resources to continuously formulate projections and comprehensive recommendations – similar to existing bodies such as the Council of Economic Experts or the Federal Chancellor’s Council on the Future.
A permanent body with a clear mandate can take a closer look at Germany’s role in the AI value chain and benefit from greater independence and guaranteed public and political attention. The existing, outstanding expertise in German industry, science and civil society should be urgently utilized.
The USA, UK, Japan, Singapore and France have each set up their own national AI safety authority (AI Safety Institute, AISI for short) since 2023. As state authorities, they research the risks of particularly advanced AI: they develop risk evaluations for AI models, continuously monitor the progress of AI and draw up regulatory proposals. The authorities cooperate closely, for example by exchanging testing methods and information on leading models.
The EU Commission’s AI Office, recently created as part of the AI Act, comes close to an AISI in parts. If the independence and adequate resources of the AI Office are ensured, it can contribute significantly to building European state capacity.
Nevertheless, Germany should not rely on this alone: Firstly, the AI Office is already responsible for enforcing the AI Regulation – a task that should not compete with advisory research work within the same organization. Secondly, the German government does not have the direct, confidential and low-threshold access to the AI Office which would be important for short-term crisis responses or strategic realignments.
There is no way around technical expertise for policy advice, reliable research and targeted policy development in the German administration. Germany could learn from the successes of international partners in establishing their AISIs, perhaps even by setting up its own German AISI, for example within the BMI or BMBF remit.
The German state lacks the ability to anticipate and navigate the technical developments and social effects of AI. This can be changed: Following the international model, the German government could endeavor to recruit technical expertise for the affected departments, convene a permanent advisory body with international experts and create its own capacities in the field of AI security.
Daniel Privitera is the founder and Executive Director of the KIRA Center, an independent think tank for AI policy. He is also Lead Writer of the International Scientific Report on the Safety of Advanced AI. Anton Leicht is a Policy Specialist at the KIRA Center, works on economic and security aspects of AI policy and is doing his PhD on democratic regulation of advanced AI systems.
Their joint report with AI policy recommendations for Germany from 2025 is published today.
It’s summit time again – this time not in Brussels, but in Frankfurt. And it is not heads of state and government who are meeting, but experts from politics, business, science and civil society for the Digital Summit 2024. The EU Commission is also present in Frankfurt.
This morning, Prabhat Agarwal, Head of the Digital Services and Platforms Unit at DG Connect, will be among those discussing the results of two years of the Digital Services Act. Hate speech and disinformation pose a challenge for politicians, especially in the super-election year 2024. Just like ahead of next year’s general election. Are the instruments of the DSA sufficient to develop the necessary clout against the major platforms? This will be one of the questions to be discussed.
However, the dominant topic at the digital summit, running until Tuesday, will be artificial intelligence: The topics range from “AI made in Europe – sovereign, trustworthy, secure” to “AI in daily business.” And, of course, the question of the correct regulation of AI will also be on the agenda. According to the experts from the KIRA Center think tank in Berlin, Germany must finally pool its expertise in this area.
Meanwhile, the EU is moving forward with the implementation of the AI Act. This provides for the establishment of a scientific committee of independent experts. The panel is to advise and support the EU AI Office and the national market surveillance authorities in the implementation and enforcement of the AI Act. To this end, the Commission has now submitted a draft implementing regulation and is now asking for comments by November 15. Perhaps you also have some ideas.
Have a good start to the week,
Anyone reading the Draghi report on telecommunications will find many of the demands that the major network operators have been making for years. The report sees an urgent need for consolidation and the promotion of European champions. This is necessary in order to achieve higher investment rates in high-performance networks. Less regulation at the national level and harmonization of spectrum licenses in mobile communications are also necessary.
In fact, the European market is fragmented, while large providers have emerged in other regions. In the EU (450 million inhabitants), there are three to four mobile network operators (MNOs) per member state. In the USA (335 million), there are only three large MNOs nationwide, as there are in China (1.4 billion). French President Emmanuel Macron summed it up in a simple formula: Bigger players can invest more, which means more innovation. However, experts doubt that this equation is correct.
The EU Commission’s Directorate-General for Competition presented its own report in the summer – and came to different conclusions: According to the report, mergers of mobile providers generally led to higher prices for consumers. Positive effects on investments in networks or the introduction of innovative technologies, on the other hand, were not clearly demonstrated. The average revenue per user (ARPU) in Europe is far below that of other regions: In 2022, Europeans spent an average of €15, compared to €42.50 in the USA, €26.50 in South Korea and €25.90 in Japan.
The analysis also shows that European countries with fewer network operators tend to have higher ARPUs. At the same time, there is a negative correlation between market concentration and investments. A larger number of network operators therefore appears to be associated with higher investments and potentially lower prices, while concentration leads to an increase in prices without significant benefits for investments or service quality.
This is in line with the observations of Tomaso Duso, Head of the Department of Enterprises and Markets at the German Institute for Economic Research (DIW Berlin) and Chairman of the Monopolies Commission. “The Draghi Report is very scientifically sound in its analysis in many areas. But the observation that market concentration leads to higher investments is not empirically proven for mobile communications markets,” says Duso. “In fact, we were able to show that joint investments at market level are higher in markets with more providers than in markets with fewer providers.”
Providers, on the other hand, argue with economies of scale. Markus Haas, CEO of o2 Telefónica, put it in simple terms at an event in Berlin: E-Plus and o2, the two smaller providers on the German market, used to invest in two different networks. Since the merger, they have become one competitive network. Instead of investing €500 million each in two networks, o2 Telefónica is now investing €1 billion in a single network.
However, in other national markets, the Commission’s competition watchdogs have not permitted such mergers. In Germany, too, the condition was that there had to be a new fourth provider (which, however, did not get off the ground). The result is an enormous fragmentation of the market.
“As a consequence, we see that the return on investment is too low for the large investments that will now be necessary over the years in various mobile and fiber optic infrastructures,” says Roman Friedrich, Partner at the consulting firm BCG. A BCG analysis shows: “Most telecommunications companies in Europe do not earn their cost of capital.”
However, the reason is complex. It could be that the price level is too low or that there really is a lack of innovation. “But initially the cost of capital is high because we had high interest rates,” says Friedrich. However, Macron’s equation goes a little further. The conclusion that innovative strength increases with size cannot necessarily be drawn from this. “Sufficient earning power may be a necessary prerequisite. But that is not enough to be innovative. It takes more than that.”
His evidence: even in countries with higher profitability, the real innovations do not come from the telecommunications companies. “The innovations come from the digital companies – these are not the existing or past champions,” he says. Duso is convinced: “The innovations come from network technology.” And in this area, Europe already has two European champions in Nokia and Ericsson.
“Of course, it is difficult to get an innovation to scale in a small market,” says Friedrich. However, larger telecommunications companies that can scale across national borders are not necessarily more innovative – not even in the USA. “Innovation culture is a corporate culture, a management characteristic and therefore independent of size.”
For Duso, cross-border mergers could indeed bring advantages. However, this would not require a revision of merger control. The existing regulations do not stand in the way of this. However, he considers it downright dangerous to facilitate non-cross-border mergers and to define the telecommunications market EU-wide for this purpose. The markets have so far been structured nationally or regionally: “The demand side can only access domestic offers locally,” explains Duso. An EU-wide market definition would therefore facilitate mergers, which would lead to higher prices and lower investments.
The competition authorities have fewer reservations about cross-border mergers per se. Some large companies such as Deutsche Telekom or Vodafone have already expanded and become important players in several member states. However, the experiences with this are mixed.
The complexity of such a transnational investment is great, says consultant Friedrich. “You simply need a lot more time and intelligence if you want to implement a scaling model in Europe than in the USA.” There are huge differences on both the capital market and regulatory side.
Friedrich can therefore understand the demand for a single internal market, as well as an EU-wide harmonization of frequency allocation, which companies would like to see. Duso also suggests promoting initiatives to create a more integrated market, for example by harmonizing frequency allocation procedures.
A market structure such as that in the USA would only be desirable if a genuine European market also existed. Before that, a simple comparison of the quality of mobile networks here with those in the USA, for example, is not possible across the board, says Duso. “There are a lot of differences in Europe. There are countries with fantastic networks, not just the Netherlands, but also Sweden and Switzerland. And there are countries with poorer networks. Unfortunately, Germany is one of these countries.” Despite Germany being the largest mobile communications market in Europe.
Mr Fooy, the Swedish Ministry of Justice created the Psychological Defense Agency (PDA) in 2022 to combat foreign false information. What was the reason for setting up this new agency?
In 2014, right before Russia’s annexation of Crimea, we noticed an increase in new forms of disinformation and propaganda. This made restarting a psychological defense capability more urgent. In 2019, the government announced the creation of a new agency for psychological defense. We can think of it as part of Sweden’s broader defense against hybrid threats.
The Swedish government pointed out that Russia, China, and Iran are countries known for disinformation campaigns targeting Sweden. What makes their approaches unique or particularly concerning?
I would say that broadly speaking, Russia has the most experience in reinforcing polarization and segmentation within society. China, of course, has the most significant resources to target vulnerabilities. And Iran is somewhat of a late bloomer. We recently had a very serious campaign against our Child protective services. Various groups in the Middle East, which are not directly state-affiliated, claim that the Swedish protective services were abducting Muslim children and trying to turn them into Christians or turn them over to be raised by homosexual parents. We and others handled this very well, but it could have led to dire consequences.
What does your authority’s work look like when confronted with such “foreign malign information influence” attacks, as you call them?
First of all: We are not an intelligence agency. We work entirely with open sources and provide reports on an operational basis to our government. Much of our work is aimed at building confidence within the Swedish population. A lot of our efforts focus on building public trust, educating people, and collaborating with social services, religious organizations, and other key institutions. However, we don’t aim to counter every piece of disinformation from Moscow, Beijing, or Tehran. That would be a superhuman effort and sometimes even counterproductive.
What challenges do you face especially when it comes to China?
We have to keep in mind that in authoritarian regimes, most information influence activities target the local population at home. That being said, however, China has vast manpower to become active on a global scale. The Chinese state media has an incredible reach. They are published in dozens of languages and are often incredibly professionally produced. I’ve seen cases where reputable outlets like the BBC unknowingly cite Chinese state media sources. Their influence is growing, especially in the Global South. For example, a Chinese publisher bought publishing rights to most of the in-flight magazines in Africa.
How does that affect Western democracies like Sweden?
As mentioned, China’s influence activities increasingly target topics beyond its borders. After the Israel-Hamas war broke out, we noticed anti-Semitic narratives in Chinese state media, for example. It’s peculiar, as anti-Semitism isn’t a common sentiment among Chinese citizens. However, these media reports can be read by Arabic speakers across the Middle East, North Africa, and also Europe, mainly without the readers realizing the source is Chinese. The same thing happens with Russian sources, and many people consuming this content might not question its origin. We want to change that.
Sometimes, it feels like China and Russia are coordinating their disinformation campaigns.
No, I wouldn’t say they coordinate. They do have agreements, like journalist exchange programs, and they amplify each other’s messaging at times. For example, narratives in Russian media may appear in Chinese outlets and vice versa. But calling it “coordination” would be an overstatement. Both are united by the perception of shared external enemies, but their operations are not fully collaborative in the Western sense.
How do you balance protecting Sweden from disinformation while maintaining freedom of speech, especially when Swedish citizens might inadvertently spread false information?
Freedom of speech is critical, and we don’t engage in censorship or fact-checking. That’s for the media and fact-checking organizations to handle. However, we focus on media literacy, ensuring that the public can critically evaluate what they’re reading and understand the intent behind certain narratives. We work with educational institutions to include media literacy in school curricula and have trained over 20,000 individuals on how to identify and counter foreign disinformation.
The PDA also prepares Sweden for the possibility of war. How would the agency support the government during a national security crisis?
While I can’t discuss wartime planning in detail, I can confirm that our mandate includes preparing for all scenarios, from crises to full-scale war. We’re part of Sweden’s total defense strategy.
I’m sure there’s much to learn from the situation in Ukraine and the Russian disinformation there.
Yes, absolutely. The situation in Ukraine is of paramount importance here, as well, because it is of great concern to our government and the Swedish population. And when it comes to figuring out how Chinese information flows works, we also look at Taiwan. But every democracy has a piece of the puzzle to help us understand how these state actors operate.
What challenges do you foresee in the future, for example, due to the advent of new technologies?
AI and deepfakes will be significant challenges, especially in terms of how much information can be flooded into the system. But we do have some advantages – Chinese disinformation, for example, is often constrained by the CCP’s rigid doctrines. Russia, on the other hand, was more imaginative during Crimea’s annexation than in its current full-scale invasion of Ukraine. The quality of disinformation varies depending on the topic and who’s behind it.
So, does that mean the quality differs significantly, even if it comes from the same state actor?
Most state actors do not have one agency dealing with information influence activities. There might be certain think tanks and security services that are part of the armed forces. There’s also a market for disinformation campaigns in China where regular nonaffiliated nationalists all of a sudden get very upset about some trope or narrative in the media space, and they do not necessarily follow the will of the country. So my greatest concern here is that somebody or a number of people in one of the authoritarian states starts being really imaginative and gets lots of resources from their governments. Such an entity could create activities that target vulnerabilities in our society that we aren’t even aware of yet.
Frederick Fooy has been working against information influence activities from various antagonistic actors for more than a decade. He has previous experience from the Swedish Armed Forces dating back to 1989, including deployments to the Balkans. Frederick has a B.A. in Byzantinology from New York University and an M.A. in Urban Policy from Brooklyn College.
The progress made in recent years has turned the former niche topic of artificial intelligence (AI) into a central aspect of economic and security policy. Many countries are responding to this with considerable determination and speed: for example, by quickly and unbureaucratically setting up AI security authorities, recruiting top technical talent and through close exchange with experts from industry and science in specially convened committees. There is currently far too little of this determination in Germany – although we urgently need it now that, among other things, a future-proof implementation of the European AI Act is pending.
The most computationally intensive general-purpose AI (GPAI) models in particular, such as GPT-4, will bring drastic changes to our society: Many experts predict potentially far-reaching effects on productivity or the education system, but also risks such as misuse. This can only be handled with care by those who develop a deep understanding of the technology and its political implications.
This understanding must extend far beyond the Federal Network Agency or a possible new digital ministry to all affected topics: Every specialist policy area must be able to assess opportunities and risks instead of being permanently reliant on outside expertise.
To achieve this, we need to draw on the existing pool of AI experts, many of whom have a German degree. Other countries are already doing this: the UK’s Frontier AI Taskforce (now the AI Safety Institute) managed to recruit top AI researchers with a total of 150 years of scientific experience in just a few months.
Competition for talent is fierce; remuneration and typical career models of the German administration can hardly keep up. Here too, other countries are more flexible: for example, research scientists at the British AI Safety Institute can earn the equivalent of more than €170,000 a year. In addition, fast tracks are being created and formal hurdles removed.
In the pioneering AI paradigm of recent years, deep learning-based general-purpose AI, Germany is only in second divison globally. Virtually all leading GPAI models are currently being developed in the USA, the UK and China. To ensure that we do not miss out on the next developments, Germany now needs bundled expertise.
The German government should therefore convene a committee of leading international experts and provide it with the necessary resources to continuously formulate projections and comprehensive recommendations – similar to existing bodies such as the Council of Economic Experts or the Federal Chancellor’s Council on the Future.
A permanent body with a clear mandate can take a closer look at Germany’s role in the AI value chain and benefit from greater independence and guaranteed public and political attention. The existing, outstanding expertise in German industry, science and civil society should be urgently utilized.
The USA, UK, Japan, Singapore and France have each set up their own national AI safety authority (AI Safety Institute, AISI for short) since 2023. As state authorities, they research the risks of particularly advanced AI: they develop risk evaluations for AI models, continuously monitor the progress of AI and draw up regulatory proposals. The authorities cooperate closely, for example by exchanging testing methods and information on leading models.
The EU Commission’s AI Office, recently created as part of the AI Act, comes close to an AISI in parts. If the independence and adequate resources of the AI Office are ensured, it can contribute significantly to building European state capacity.
Nevertheless, Germany should not rely on this alone: Firstly, the AI Office is already responsible for enforcing the AI Regulation – a task that should not compete with advisory research work within the same organization. Secondly, the German government does not have the direct, confidential and low-threshold access to the AI Office which would be important for short-term crisis responses or strategic realignments.
There is no way around technical expertise for policy advice, reliable research and targeted policy development in the German administration. Germany could learn from the successes of international partners in establishing their AISIs, perhaps even by setting up its own German AISI, for example within the BMI or BMBF remit.
The German state lacks the ability to anticipate and navigate the technical developments and social effects of AI. This can be changed: Following the international model, the German government could endeavor to recruit technical expertise for the affected departments, convene a permanent advisory body with international experts and create its own capacities in the field of AI security.
Daniel Privitera is the founder and Executive Director of the KIRA Center, an independent think tank for AI policy. He is also Lead Writer of the International Scientific Report on the Safety of Advanced AI. Anton Leicht is a Policy Specialist at the KIRA Center, works on economic and security aspects of AI policy and is doing his PhD on democratic regulation of advanced AI systems.
Their joint report with AI policy recommendations for Germany from 2025 is published today.