In an unpredictable world, howresilient is your infrastructure?

On 8 September, Romania's intelligence service announced that it had prevented a Russian-coordinated sabotage operation involving surveillance of military facilities and Ukrainian cargo aircraft. A week earlier, Germany attributed August's attempted explosive-drone attack at Leipzig/Halle airport to Russia. These cases demand operational deterrence and defence alongside resilience against physical hybrid threats. At the 15–16 October European Council, leaders should agree to pool protective capabilities and disrupt enabling networks, backed by a protocol under the EU Hybrid Toolbox for rapid assessment and response. The objective should be deterrence by dismantlement: reducing adversaries' capacity to organise and repeat attacks.

Resilience needs operational protection

Europe has strengthened its response to disinformation, propaganda and election interference through the Foreign Information Manipulation and Interference Toolbox and closer coordination. It has also adopted cable-security and counter-drone plans. The next step is to connect these commitments to readily available capabilities, overcoming the institutional divisions between intelligence, infrastructure protection and law enforcement.

The distinction between digital and physical threats is increasingly misleading. An August investigation disclosed that December's cyberattack on a Polish heat-and-power plant had stopped a steam turbine. Operators prevented disruption to household heating. That successful response demonstrates the value of resilience, but also the need to prevent hostile access from becoming physical disruption.

Protecting information spaces cannot stop an explosive drone or defend a cable landing station. Nor can authorities rely indefinitely on expensive interceptors against inexpensive drones. Ukraine's experience points towards layered protection using affordable systems that can be upgraded rapidly. Protection must extend to ports, energy installations and transport hubs inside Europe, not only its external borders.

Sabotage can force Europeans to divert resources, undermine confidence in essential services and complicate support for Ukraine. A damaged connection may be repaired quickly; persistent uncertainty about whether the next attack can be stopped carries a broader political cost. Repeated disruption can become coercion without approaching the threshold of conventional war.

Dismantle the means of attack

Deterrence by dismantlement would connect this protective effort to sustained action against the people, financing and logistics enabling attacks. It complements deterrence through denial and punishment; it does not replace them. Its contribution is to make disruption cumulative: each investigation should leave an adversary with fewer usable networks and resources.

There is a tested foundation. In March, a joint investigation supported by Eurojust and Europol identified 22 suspects in a cross-border parcel-arson campaign. Such cooperation can connect offences that appear isolated nationally. However, arresting expendable recruits will have limited deterrent value if recruiters, payment channels and logistical intermediaries remain available. Investigations should therefore systematically pursue the enabling network, not conclude with the immediate perpetrator.

There are legitimate objections. National authorities control security operations; intelligence cannot always be disclosed; premature attribution risks error and escalation. A common protocol should not lower evidential standards or transfer command to Brussels. Protective deployments need not await public attribution, while coercive measures require a sound legal basis and review. NATO should retain its collective-defence role. The alternative, fragmented national responses, leaves adversaries opportunities to relocate their activities and impose repeated disruption on European services and support for Ukraine.

Give dismantlement an operational timetable

October's European Council should set three tasks.

First, the High Representative and Commission should present the protocol by December, implementing Council's March conclusions on hybrid threats. Following a serious suspected physical hybrid attack, the EU Hybrid Fusion Cell should aim to provide a preliminary assessment within 24 hours, drawing on national reports and stating uncertainties. Within 72 hours, ministers should receive response options identifying the responsible authority, legal basis and available resources. Immediate protection must not wait for this process.

Second, the Commission's plan to explore rapid counter-drone teams by year-end should produce national commitments to a deployable pool by June 2027. Member States should earmark crews, mobile sensors and affordable interception systems, agreeing assistance requests, deployment permissions and command arrangements beforehand. The joint European drone and seabed defence projects proposed in July should prioritise these capability gaps, including surveillance and inspection equipment around critical cables. Member States should finance initial standby capacity, using the European Defence Industry Programme for eligible joint capability development. Funding must cover maintenance and regular upgrades, informed by Ukraine's experience.

Third, national prosecutors should use Eurojust-supported joint investigations whenever evidence links attacks across borders, with Europol tracing associated networks. The Council should promptly assess Germany's proposed hybrid-sanctions listings. Member States should pair any listings with investigations into the organisers' financial and logistical support, pursuing criminal asset seizures where the evidence permits. By June 2027, ministers should review deployment times, protective equipment actually available and networks disrupted. A joint exercise with NATO and infrastructure operators should test the protocol across drone, cable and energy incidents.

Inaction would leave Europe paying repeatedly to restore services while hostile networks reconstitute. October should mark a shift from commitments to capabilities. Europe's answer to the next attack should leave fewer means for the one after it.

On 8 September, Romania's intelligence service announced that it had prevented a Russian-coordinated sabotage operation involving surveillance of military facilities and Ukrainian cargo aircraft. A week earlier, Germany attributed August's attempted explosive-drone attack at Leipzig/Halle airport to Russia. These cases demand operational deterrence and defence alongside resilience against physical hybrid threats. At the 15–16 October European Council, leaders should agree to pool protective capabilities and disrupt enabling networks, backed by a protocol under the EU Hybrid Toolbox for rapid assessment and response. The objective should be deterrence by dismantlement: reducing adversaries' capacity to organise and repeat attacks.

Resilience needs operational protection

Europe has strengthened its response to disinformation, propaganda and election interference through the Foreign Information Manipulation and Interference Toolbox and closer coordination. It has also adopted cable-security and counter-drone plans. The next step is to connect these commitments to readily available capabilities, overcoming the institutional divisions between intelligence, infrastructure protection and law enforcement.

The distinction between digital and physical threats is increasingly misleading. An August investigation disclosed that December's cyberattack on a Polish heat-and-power plant had stopped a steam turbine. Operators prevented disruption to household heating. That successful response demonstrates the value of resilience, but also the need to prevent hostile access from becoming physical disruption.

Protecting information spaces cannot stop an explosive drone or defend a cable landing station. Nor can authorities rely indefinitely on expensive interceptors against inexpensive drones. Ukraine's experience points towards layered protection using affordable systems that can be upgraded rapidly. Protection must extend to ports, energy installations and transport hubs inside Europe, not only its external borders.

Sabotage can force Europeans to divert resources, undermine confidence in essential services and complicate support for Ukraine. A damaged connection may be repaired quickly; persistent uncertainty about whether the next attack can be stopped carries a broader political cost. Repeated disruption can become coercion without approaching the threshold of conventional war.

Dismantle the means of attack

Deterrence by dismantlement would connect this protective effort to sustained action against the people, financing and logistics enabling attacks. It complements deterrence through denial and punishment; it does not replace them. Its contribution is to make disruption cumulative: each investigation should leave an adversary with fewer usable networks and resources.

There is a tested foundation. In March, a joint investigation supported by Eurojust and Europol identified 22 suspects in a cross-border parcel-arson campaign. Such cooperation can connect offences that appear isolated nationally. However, arresting expendable recruits will have limited deterrent value if recruiters, payment channels and logistical intermediaries remain available. Investigations should therefore systematically pursue the enabling network, not conclude with the immediate perpetrator.

There are legitimate objections. National authorities control security operations; intelligence cannot always be disclosed; premature attribution risks error and escalation. A common protocol should not lower evidential standards or transfer command to Brussels. Protective deployments need not await public attribution, while coercive measures require a sound legal basis and review. NATO should retain its collective-defence role. The alternative, fragmented national responses, leaves adversaries opportunities to relocate their activities and impose repeated disruption on European services and support for Ukraine.

Give dismantlement an operational timetable

October's European Council should set three tasks.

First, the High Representative and Commission should present the protocol by December, implementing Council's March conclusions on hybrid threats. Following a serious suspected physical hybrid attack, the EU Hybrid Fusion Cell should aim to provide a preliminary assessment within 24 hours, drawing on national reports and stating uncertainties. Within 72 hours, ministers should receive response options identifying the responsible authority, legal basis and available resources. Immediate protection must not wait for this process.

Second, the Commission's plan to explore rapid counter-drone teams by year-end should produce national commitments to a deployable pool by June 2027. Member States should earmark crews, mobile sensors and affordable interception systems, agreeing assistance requests, deployment permissions and command arrangements beforehand. The joint European drone and seabed defence projects proposed in July should prioritise these capability gaps, including surveillance and inspection equipment around critical cables. Member States should finance initial standby capacity, using the European Defence Industry Programme for eligible joint capability development. Funding must cover maintenance and regular upgrades, informed by Ukraine's experience.

Third, national prosecutors should use Eurojust-supported joint investigations whenever evidence links attacks across borders, with Europol tracing associated networks. The Council should promptly assess Germany's proposed hybrid-sanctions listings. Member States should pair any listings with investigations into the organisers' financial and logistical support, pursuing criminal asset seizures where the evidence permits. By June 2027, ministers should review deployment times, protective equipment actually available and networks disrupted. A joint exercise with NATO and infrastructure operators should test the protocol across drone, cable and energy incidents.

Inaction would leave Europe paying repeatedly to restore services while hostile networks reconstitute. October should mark a shift from commitments to capabilities. Europe's answer to the next attack should leave fewer means for the one after it.

Experts defining the activities of the Russian Orthodox Church (ROC) as part of Russia's hybrid warfare (HW) often focus solely on the ROC's propaganda against Ukraine. This leads to an incomplete account of how Russia weaponizes religion. Conceptually, HW refers not merely to disinformation but to a combination of conventional and unconventional warfighting methods. More importantly, empirical evidence indicates that elements of the ROC act not only as a disinformation machine but also as a troop recruitment platform, including in Africa. The recent expansion of the ROC into Africa is critical to Western security since, in some African states, the ROC is reportedly recruiting irregular soldiers to join the war against Ukraine. To offer an updated research and policy agenda, we examine the effectiveness of the ROC's hybrid operations in the Sahel. This case has broader methodological relevance for studying the ROC's activities across Africa. The aim is to lay the foundations for a critical academic discussion of the ROC's HW activities in Africa and to derive best practices practitioners can use to evaluate HW effectiveness.

By examining various open-source empirical data, we argue that it is analytically useful to define the ROC's malign activities in the Sahel region as hybrid operations. These operations are thus far effective, producing strategic, operational, and tactical effects in support of Russia's HW against the West. We conclude that a successful HW signature has seven basic elements: adaptability, ambiguity, asymmetry, deniability, irregularity, multimodality, and synchronicity. The West must support African states in defending against such hybrid operations.

Experts defining the activities of the Russian Orthodox Church (ROC) as part of Russia's hybrid warfare (HW) often focus solely on the ROC's propaganda against Ukraine. This leads to an incomplete account of how Russia weaponizes religion. Conceptually, HW refers not merely to disinformation but to a combination of conventional and unconventional warfighting methods. More importantly, empirical evidence indicates that elements of the ROC act not only as a disinformation machine but also as a troop recruitment platform, including in Africa. The recent expansion of the ROC into Africa is critical to Western security since, in some African states, the ROC is reportedly recruiting irregular soldiers to join the war against Ukraine. To offer an updated research and policy agenda, we examine the effectiveness of the ROC's hybrid operations in the Sahel. This case has broader methodological relevance for studying the ROC's activities across Africa. The aim is to lay the foundations for a critical academic discussion of the ROC's HW activities in Africa and to derive best practices practitioners can use to evaluate HW effectiveness.

By examining various open-source empirical data, we argue that it is analytically useful to define the ROC's malign activities in the Sahel region as hybrid operations. These operations are thus far effective, producing strategic, operational, and tactical effects in support of Russia's HW against the West. We conclude that a successful HW signature has seven basic elements: adaptability, ambiguity, asymmetry, deniability, irregularity, multimodality, and synchronicity. The West must support African states in defending against such hybrid operations.

On 18 September, CNN reported that US forces had prepared to intercept a Chinese vessel earlier this year after AI-assisted intelligence falsely identified its cargo as nuclear weapons programme components. At their 15–16 October meeting, EU leaders should launch a joint programme securing the entire AI value chain, from critical materials to tested military applications, through coordinated investment and common procurement, so that European forces can use reliable AI and sustain it independently.

Anthropic's 10 September report reinforces the urgency. It describes Claude's use in suspected Russian state-linked cyberespionage targeting Ukraine, weapons research by actors assessed as linked to China's defence sector, and drone-swarm software development by Russian-based freelancers. These findings do not establish that all these projects became operational military capabilities. They nevertheless show commercial AI entering both hostile operations and military development. Europe needs the ability to develop, verify and maintain its own systems.

Secure the material foundations

Europe has invested in supercomputers, semiconductor capacity and defence research. The Commission's June technology sovereignty package recognises the remaining gaps. Yet a European model running on imported processors, foreign-controlled software and vulnerable electricity supplies still leaves military readiness exposed. AI autonomy is a condition for strategic autonomy because external restrictions can determine whether European forces can operate and adapt their equipment.

That dependence begins underground. Rare earths support magnets in defence equipment; gallium and germanium matter for specialised electronics and sensors. Securing AI therefore requires attention to its wider industrial and energy foundations. China's mineral export restrictions already demonstrate how supply concentration can become strategic pressure. The European Court of Auditors found that efforts to diversify critical raw material imports for the energy transition had not delivered tangible results.

The Commission and participating governments should turn strategic mineral partnerships into long-term purchase commitments, supporting European and partner-country refining, recycling and component production. Guaranteed demand would help alternative suppliers attract capital despite higher initial costs, while competitive tenders would limit subsidies and protect taxpayers. Targeted stockpiles should bridge supply interruptions. During negotiations on Chips Act 2.0, legislators should prioritise defence-relevant gaps in chip design, manufacturing and advanced packaging. Funding should be tied to delivery milestones and verified supply resilience.

Turn computing into military capability

As EUISS analysis argues, fragmented digital infrastructure limits Europe's defence capabilities. The proposed Cloud and AI Development Act offers an opening to align civilian investment with defence requirements. Member States should commission accredited computing capacity for sensitive military workloads, backed by resilient electricity, protected communications and independently tested cybersecurity. Civilian computing expansion alone does not guarantee secure military access.

The European Defence Agency should assemble willing governments around common requirements and purchasing commitments, beginning with logistics, maintenance and intelligence analysis. Contracts should require interoperable data formats, access to the software and model components needed for maintenance, and enforceable rights to operate and update systems if a supplier withdraws. Ministries should agree rules for sharing releasable training and evaluation data, preserving national control over classified information. Smaller specialised models can serve many missions without reproducing every frontier AI capability.

EU defence research funding should connect successful prototypes to these purchases. Contracts should finance continuous updates, testing and specialist training. Cooperation with Ukrainian developers could bring operational experience into joint development while strengthening Ukraine's industrial capacity. National authorities would retain deployment decisions; the EU would organise the industrial conditions for a stronger European contribution to NATO.

Keep missions under accountable control

European ownership cannot prevent erroneous intelligence. The AI Act excludes systems used exclusively for military, defence or national security purposes. Participating defence ministries should therefore make procurement conditional on common assurance requirements: traceable evidence, independent testing against manipulated data, recorded human decisions and effective authority to halt unsafe use. Analysts must be trained to challenge outputs and independently corroborate consequential claims. A human signature cannot substitute for scrutiny.

The strongest objection is that duplicating global supply chains would absorb scarce defence resources and slow adoption. That concern should shape priorities. Europe should retain diversified allied suppliers while building domestic capacity where interruption would prevent critical missions. Joint demand can make those investments viable and reduce national duplication. Inaction also carries costs: interrupted access, dependence on suppliers' decisions and growing difficulty replacing systems embedded in military operations.

The October European Council should request a costed implementation plan from the Commission and European Defence Agency within six months, with participating governments naming initial joint purchases. National procurement budgets and eligible projects under existing EU defence instruments should support initial trials in 2027. Negotiations on the 2028–2034 budget should provide sustained funding, with annual tests of continued operation and recovery after supplier, power or network disruption. Progress should be judged by capabilities delivered and dependencies reduced. Europe's freedom to act must extend from minerals to missions.

On 18 September, CNN reported that US forces had prepared to intercept a Chinese vessel earlier this year after AI-assisted intelligence falsely identified its cargo as nuclear weapons programme components. At their 15–16 October meeting, EU leaders should launch a joint programme securing the entire AI value chain, from critical materials to tested military applications, through coordinated investment and common procurement, so that European forces can use reliable AI and sustain it independently.

Anthropic's 10 September report reinforces the urgency. It describes Claude's use in suspected Russian state-linked cyberespionage targeting Ukraine, weapons research by actors assessed as linked to China's defence sector, and drone-swarm software development by Russian-based freelancers. These findings do not establish that all these projects became operational military capabilities. They nevertheless show commercial AI entering both hostile operations and military development. Europe needs the ability to develop, verify and maintain its own systems.

Secure the material foundations

Europe has invested in supercomputers, semiconductor capacity and defence research. The Commission's June technology sovereignty package recognises the remaining gaps. Yet a European model running on imported processors, foreign-controlled software and vulnerable electricity supplies still leaves military readiness exposed. AI autonomy is a condition for strategic autonomy because external restrictions can determine whether European forces can operate and adapt their equipment.

That dependence begins underground. Rare earths support magnets in defence equipment; gallium and germanium matter for specialised electronics and sensors. Securing AI therefore requires attention to its wider industrial and energy foundations. China's mineral export restrictions already demonstrate how supply concentration can become strategic pressure. The European Court of Auditors found that efforts to diversify critical raw material imports for the energy transition had not delivered tangible results.

The Commission and participating governments should turn strategic mineral partnerships into long-term purchase commitments, supporting European and partner-country refining, recycling and component production. Guaranteed demand would help alternative suppliers attract capital despite higher initial costs, while competitive tenders would limit subsidies and protect taxpayers. Targeted stockpiles should bridge supply interruptions. During negotiations on Chips Act 2.0, legislators should prioritise defence-relevant gaps in chip design, manufacturing and advanced packaging. Funding should be tied to delivery milestones and verified supply resilience.

Turn computing into military capability

As EUISS analysis argues, fragmented digital infrastructure limits Europe's defence capabilities. The proposed Cloud and AI Development Act offers an opening to align civilian investment with defence requirements. Member States should commission accredited computing capacity for sensitive military workloads, backed by resilient electricity, protected communications and independently tested cybersecurity. Civilian computing expansion alone does not guarantee secure military access.

The European Defence Agency should assemble willing governments around common requirements and purchasing commitments, beginning with logistics, maintenance and intelligence analysis. Contracts should require interoperable data formats, access to the software and model components needed for maintenance, and enforceable rights to operate and update systems if a supplier withdraws. Ministries should agree rules for sharing releasable training and evaluation data, preserving national control over classified information. Smaller specialised models can serve many missions without reproducing every frontier AI capability.

EU defence research funding should connect successful prototypes to these purchases. Contracts should finance continuous updates, testing and specialist training. Cooperation with Ukrainian developers could bring operational experience into joint development while strengthening Ukraine's industrial capacity. National authorities would retain deployment decisions; the EU would organise the industrial conditions for a stronger European contribution to NATO.

Keep missions under accountable control

European ownership cannot prevent erroneous intelligence. The AI Act excludes systems used exclusively for military, defence or national security purposes. Participating defence ministries should therefore make procurement conditional on common assurance requirements: traceable evidence, independent testing against manipulated data, recorded human decisions and effective authority to halt unsafe use. Analysts must be trained to challenge outputs and independently corroborate consequential claims. A human signature cannot substitute for scrutiny.

The strongest objection is that duplicating global supply chains would absorb scarce defence resources and slow adoption. That concern should shape priorities. Europe should retain diversified allied suppliers while building domestic capacity where interruption would prevent critical missions. Joint demand can make those investments viable and reduce national duplication. Inaction also carries costs: interrupted access, dependence on suppliers' decisions and growing difficulty replacing systems embedded in military operations.

The October European Council should request a costed implementation plan from the Commission and European Defence Agency within six months, with participating governments naming initial joint purchases. National procurement budgets and eligible projects under existing EU defence instruments should support initial trials in 2027. Negotiations on the 2028–2034 budget should provide sustained funding, with annual tests of continued operation and recovery after supplier, power or network disruption. Progress should be judged by capabilities delivered and dependencies reduced. Europe's freedom to act must extend from minerals to missions.