The Geopod about AI and geopolitics
The conversation in the attached podcast with Andreas von der Heide and Lars-Erik Lundin tests the thesis that artificial intelligence simultaneously reinforces threats to the rules-based world order and constitutes a tool needed to meet them.
Thesis: AI in an era of simultaneous pressures
The starting point is that the climate, the UN order and the security architecture are under pressure when the systems' ability to absorb complexity no longer exceeds the cost. However, “collapse” is a rhetorical device – it is as much about transformation, where AI both accelerates the breakdown and offers tools to manage it.
The threat image: disorientation, not just lies
The Russian disinformation campaign Doppelgänger, run by the Social Design Agency and mapped by the EU DisinfoLab, has cloned established titles such as Der Spiegel, Bild, FAZ, Die Welt, Le Monde, Le Parisien and Le Figaro to lead readers to manipulated copies. Deepfake technology opens the way for even military doctrines to be replaced by fakes – but each concrete example, such as an allegedly genuine video clip of a crying British Prime Minister, must be independently examined critically before being relied upon. The Psychological Defense Agency’s mission is linked to Alexander L. George’s method classic Propaganda Analysis (1959), which deals with inferential content analysis of Nazi Germany's propaganda and remains relevant as a methodological foundation.
Flow control as new geopolitics
Geopolitics is increasingly about the control of flows – physical, digital, financial – rather than territory. The importance of Greenland for submarine passages, the GIUK gap, the 1970s seabed treaty, and the EU’s ISPRA container tracking and “Intelligent Border Management” illustrate this shift. The role of AI is to allow systems to absorb significantly more complexity and thus monitor flows more quickly and effectively – making technology a central resource of power.
Defensive becomes offensive – with reservations
The tennis racket metaphor formulates a “fatal strategic mistake” in defending itself only against the information operations of autocrats. Sweden should, according to the reasoning, “play tennis doubles” with Ukraine, whose 35-year-old defense minister and collaboration with Silicon Valley on Nomad drones and AI filtering of reconnaissance data make the country a teacher rather than a mere recipient. But an offensive line must be justified against the boundary conditions of the rule of law: what is legitimate for a country at war is not automatically transferable to Sweden in formal peacetime.
Europe's lag
A recurring problem is that European and Swedish decision-makers lack both curiosity and competence about the technology they lead. The Internet was not considered a priority in the government for a long time, and today there is a similar shame culture around AI: journalists often acknowledge use in conversations but rarely mention this in their texts, while American universities are introducing strict accounting requirements for research. Vice President JD Vance's Munich criticism of European regulation, the acquisitions of European AI startups and the difficulty of matching American salaries reinforce the image of a challenged continent.
The three resources of the race – with objection
Three resources are highlighted: training data, energy and hardware chains. China's 1,4 billion people constitute a data volume without GDPR restrictions; Eric Schmidt already spoke a year ago about hundreds of nuclear power plants for model training; ASML, Taiwanese chip manufacturing and Nvidia form a vulnerable chain. But DeepSeek showed that radical efficiency can drastically reduce resource requirements – which nuances both the energy and data volume arguments.
The researcher's toolbox
AI can be used for source acquisition, structuring and systematization, which frees up time for analysis. A Trump analysis on this website went via AI from around fifty scientific articles to primary sources such as Project 2025 (800 pages), the campaign website “47” and US security strategies – a meta-analysis of this material would be practically impossible without AI support. Own quality-assured databases are consulted before open internet, and one model reviews another like a “devil’s advocate”. The warning about the limitations of free versions is clear – in-depth review may otherwise be impossible.
AGI and existential threats
That AGI – general intelligence – threatens Russia, which risks not keeping up. The international debate is dominated by alignment problems, control issues and the concentration of power among a few American and Chinese actors.
Final reflection
Used correctly, AI gives time to think, to focus on the “last five percent” that is not easily available online, and to perform meta-analyses of competing explanatory models. The key is that AI draws conclusions that need to be critically examined – this is where humans must spend their time, and where European security policy research can gain a qualitative edge.