Emergencies in the U.S. require distinguishing between facts and hypotheses
In Peoria, Illinois, the discovery of an explosive device at BioUrja Renewables, a facility for the production of ethanol and industrial alcohol, introduces an element of physical threat whose strategic significance must be assessed by distinguishing the nature of the target from the intentions of whoever placed the device. Police confirmed that the device had been secured on the evening of November 13, while Police Chief Brad Dixon reported that the search was continuing and that the one-mile evacuation order would remain in effect through Saturday. Although the site’s industrial purpose requires consideration of potential industrial and regional consequences, the available information does not yet allow for determining the motive or whether the incident is part of a broader scheme; similarly, the search for additional devices is an investigative activity from which their presence cannot be inferred.
In New Orleans, the collision between the tanker Atlantic Asphalt and the river cruise ship American Heritage—which was moored and capsized—raises questions regarding the loss of control over a complex system and the reconstruction of the sequence of events leading up to it. Authorities reported that all persons on board had been accounted for and that 32 people had been hospitalized with non-life-threatening injuries, while the Coast Guard indicated an apparent loss of steering control of the tanker, the cause of which remains unknown. From an analytical perspective, this indication describes an observed operational condition without identifying its causal mechanism: to determine whether mechanical, electrical, or IT components were involved, it is necessary to reconstruct the relationship between the commands issued, the systems’ responses, and the vessel’s actual conditions, while ensuring that the geopolitical context is not used to fill in gaps that technical evidence has not yet clarified.
In the civil aviation sector, the nationwide suspension of departures ordered by the Federal Aviation Administration highlights how a communications problem can have far-reaching consequences across a critical national function. The Department of Transportation preliminarily attributed the disruption to the overload of older-generation equipment, while the Department of Homeland Security stated that no evidence of a cyberattack had emerged. Although the suspension was lifted on the evening of November 13, it continued to affect traffic management, to the extent that, according to an ABC update at 6:52 a.m. on the East Coast (prior to 3:50 p.m. Italian time), cancellations between Friday and Saturday exceeded 12,500, according to FlightAware. This strategic data therefore also pertains to the gap between the restoration of technical availability and the recovery of operational capacity, during which constraints persist that can affect users, businesses, and activities dependent on transportation.
Taken as a whole, these incidents allow for an examination of the convergence of consequences, while the hypothesis of a coordinated campaign would require verifiable links that temporal proximity alone does not provide. Cybergeopolitical analysis must distinguish between attributing causes and assessing the vulnerabilities that a crisis reveals, since an accidental disruption can expose dependencies susceptible to subsequent exploitation, even if an opportunistic actor was not the one responsible for the disruption. For Italy, the issue is understanding whether the simultaneous pressure on transportation, communications, and industrial activities could limit the options available to decision-makers, especially when recovery requires the same specialists, suppliers, or management tools. It is this potential reduction in operational autonomy—which must be assessed within national systems—that gives the comparison with the United States strategic value even before the determination of individual responsibilities is complete.
The pressure may shift to the conditions for the restart
An operational link emerges from the communication in which the National Transportation Safety Board indicated that lingering air travel delays would have forced the investigative team headed to New Orleans to travel by land. This information pertains to the organization of the investigation and does not demonstrate delays in the initial rescue efforts, but it highlights a dependency through which a disruption can alter the resources available to address a separate event. Transposed to the digital realm, this relationship requires verifying whether an operator retains the ability to intervene when the environment managing the service becomes unavailable. If privileged identities, configurations, and logs depend on that same infrastructure, even a functioning backup might remain unusable, because both the ability to activate it and the evidence needed to assess its status would be lost. The critical point would then concern the conditions under which a resumption can be authorized. An adversary who recognizes this dependency might attempt to prolong an outage that was originally accidental, without having caused it or sharing the objectives of other actors who intervened during the crisis. Attribution should therefore distinguish the initial event from subsequent intrusions and from communications intended to influence the response, since a technical explanation of the former would not, on its own, rule out subsequent hostile conduct.
To verify the dependency of the restart on these resources, exercises should identify which factors are preventing recovery, distinguishing between unavailable equipment, missing access points, and necessary checks, without adding up overlapping times. The threshold must be set before the test, in relation to the minimum service required for the population, so that the delay can be assessed based on the expected consequences. The significance of this dependency would increase if the service remained suspended despite the availability of technical resources, while it would decrease if independent alternatives allowed for a restart within the required timeframe without compromising physical safety.
Europe already provides insight into the control of physical processes
The Italian case of October 16, 2026 helps clarify the relationship between safety and operational capacity. ENAV reported that, following anomalies in the data network, emergency systems had maintained normal capacity until 6:00 a.m., while the increase in traffic subsequently required restrictions; the company, which ruled out malicious activity, later announced full recovery later that day. A reserve can therefore fulfill its protective function even while handling a volume lower than the ordinary one. The European statement of November 4 on refrigeration incidents—reported since October 11 in several Member States and followed by a new wave on November 1 and 2—offers, in contrast, a directly IT-related perspective. While calling for coordination and mitigation measures, the document places the incidents within an ongoing investigation, which does not allow for state-level attributions nor does it permit the assumption that the characteristics of the affected facilities are also present in Italian ones. On the same day, CSIRT Italia urged national entities to protect remote control systems and cold chain management interfaces, noting that unauthorized access can alter parameters, thresholds, and notifications. Since even displayed data and alarms can be modified, the bulletin recommends verifying their integrity and functionality, though this precautionary advisory does not constitute confirmation of any Italian victims. The strategic implication concerns the quality of the information upon which process control depends, since a system that responds to commands can still be operated based on an erroneous representation of its own status. For Italian officials, verifying the correspondence between a command, the received indication, and the physical effect thus becomes part of protecting workers, users, and products—especially when the urgency to resume operations reduces the time available to detect a discrepancy.
Digital forensics must make the sequence of the crisis verifiable
Forensic reconstruction first requires assessing how reliable the recorded timestamps are, since misaligned clocks and delayed recordings can alter the apparent order of events. NIST addresses these limitations, while ENISA emphasizes synchronization and the reliability of monitoring. If the potential relative error exceeds the time interval between two records, these records alone are insufficient to establish the sequence of events, making it necessary to preserve the original timestamps and document any normalization applied to the reconstruction. In industrial facilities, comparison with the log of physical operations also allows for assessing whether an anomaly stems from a manual operation or an authorized modification—which computer data alone might not explain. The NIST Operational Technology (OT) Framework brings together analysts and plant technicians and provides for monitored recovery while the investigation continues; response actions—which in turn modify the system under examination—must also remain identifiable within this process.
Comparing cryptographic fingerprints, or hashes, allows one to verify whether data matches a reference, the origin of which must be ascertained; such a match does not prove that the reference predates a compromise or describes a secure configuration for the physical process. Similarly, the recording of a command requires verification of the resulting effect, so that the preservation of intact data is not confused with the verification of the operational condition to which the data refers. This is a crucial distinction when the evidence must support the system’s restart.
For Italy, developing this capability means identifying the essential evidence in advance and making it accessible even if the normal operating environment fails, without delaying the actions necessary to protect people. Since a failure may also affect monitoring systems, the absence of alarms remains a fact that must be explained, while a limited and supervised resumption of operations may be evaluated—under defined responsibilities—before the investigation into all causes is concluded. Any remaining uncertainties should be explicitly stated in the decision, distinguishing between those that prevent recovery and those that can be managed through service limitations and additional controls.
The availability of a copy does not exhaust cybersecurity sovereignty
The June 2026 NIST guide on industrial backups notes that recovery may require compatible control logic, configurations, licenses, and components, in addition to the stored data; the Cybersecurity Framework 2.0 lists verifying the integrity of recovered resources and systems among the outcomes to be pursued. For an Italian operator, this implies a test in which the absence of the provider or the administrative environment allows for determining who can actually use the available copies. Independence must be assessed relative to the incident being mitigated, since copies located in different sites may share credentials, while separate providers may depend on the same authentication service. It is therefore necessary to verify what minimum functionality the alternative supports and for how long, recognizing that reduced capacity—provided it is managed and communicated—can preserve the protection of individuals and prevent the disruption from extending to activities that depend on that service. The Italian Cloud Strategy links data classification, service qualification, and national infrastructure to the strengthening of public administration. The geopolitical dimension emerges when multiple countries request the same intervention and a shared provider’s priorities influence the order of service resumption, without contractual availability automatically guaranteeing priority. Italy should therefore negotiate and verify simultaneous capabilities and support criteria, while maintaining alternatives for essential national functions. The issue also arises in the absence of hostilities, when a single provider’s decision—guided by its own resources and obligations—does not align with the public priority assigned to an Italian service. The autonomy sought concerns the ability to make effective choices, which can be strengthened by Italian expertise and reliable agreements with allies without requiring every component to be of national origin. When a dependency undermines the ability to intervene, its strategic importance stems from the consequences for the community, while the expertise of maintenance personnel and process managers contributes to infrastructure control in ways that the mere purchase of technology cannot ensure.
With AI, verification can become the decisive constraint
In the annual review of November 3, 2026, the British National Cyber Security Centre describes AI as a factor accelerating cyber activities, noting uneven adoption and the enduring human role in directing and validating operations. Since an identified vulnerability is not necessarily exploitable, the advantage also depends on the capabilities of those who employ the tools. For Italy, it must therefore be assessed whether the emergence of problems exceeds the capacity to examine and correct them.
The Anthropic report from September 2026 describes cases observed between December 2025 and August 2026 in which AI facilitated hostile activities and the adaptation of tools, with human intervention on key decisions; the case studies, selected by the provider, do not represent the overall incidence of the phenomenon. In terms of attribution, the increased accessibility of certain capabilities may reduce the discriminatory value of technical sophistication alone, while the model developer, infrastructure operator, and client of the operation remain distinct entities. If reports and requests for intervention exceed validation capacity, the constraint could shift to the time required to determine which actions are justified, although defenders may also use AI to improve the selection process. The verification should therefore focus on the unresolved issues that are preventing recovery, distinguishing hostile pressure from legitimate overload and organizational dysfunction, so that any slowdown is not automatically attributed to the adversary. An agent involved in the response can compare logs and propose priorities, provided that observations, inferences, and actions taken remain distinguishable, since the linguistic coherence of an explanation does not guarantee its causal correctness. Even agreement among assistants consulting the same altered source would leave the problem of corroboration unresolved, making it necessary to verify the source’s origin and to suspend automation before erroneous instructions affect privileged access or industrial processes.
AI’s offensive capabilities can enhance a measurable defense
The defensive use of AI’s offensive capabilities becomes valuable when it allows a verified weakness to be transformed into an effective correction, within an authorized and controllable scope. The NCSC envisions greater continuity between discovery, prioritization, remediation, and verification, with clear safeguards and accountability. For Italy, the criterion should focus on the time required to resolve a confirmed issue and service continuity, since an increase in reported anomalies can also complicate the verification work of specialists.
Exercises should be conducted on appropriate replicas separate from production environments, verifying whether an unavailable identity prevents access to the backup or whether unauthenticated commands bypass controls. Comparison with equivalent tests conducted without AI support would allow for the measurement of confirmed issues, successful corrections, and the human effort required, taking into account unintended outages and the limits within which a laboratory result can be extrapolated to real-world system conditions.
The British A.I. Security Institute documented unauthorized actions by certain agents against real subjects, detected on July 28, 2026, during cyber assessments, in a configuration where the Internet was accessible and certain filters had been disabled. The incident did not constitute an escape from the test environment, and the institute found no actual damage resulting from the attempts; however, it highlights why defensive intent cannot replace technical containment, without authorizing generalizations about the routine use of these models. In Italian facilities, every test must be coordinated with those who manage the physical process, ensuring restricted access, documentation of authorizations, and safe termination of operations. The goal is to test defenses and vendors by simulating hostile behavior on authorized systems, while maintaining human oversight over operational changes; a response targeting external infrastructure would involve different prerequisites and responsibilities, which do not stem from the need to improve one’s own preparedness.
Over the next three months, the authenticity of instructions must be safeguarded
Between November 2026 and mid-February 2027, one scenario to test involves the insertion of inauthentic instructions into the channels used during recovery. If a service disruption makes the need for assistance public and forces the use of alternative contacts, an opportunistic actor could pose as a maintenance technician, using available information and messages adapted to the context—possibly with the aid of AI—without having caused the initial disruption. The conditional prediction is that, in the absence of an independent verification process, even rejected hostile requests could prolong the outage, taking time away from critical verifications when a small number of decision-makers hold all the authorizations. This assessment would be strengthened by observing inauthentic instructions and delays attributable to their evaluation, while it would lose credibility if such requests were eliminated without affecting critical operations. The quality of the messages, on its own, would not prove the use of AI.
A test conducted within ninety days should compare the same outage with and without inauthentic requests, involving operators, maintenance personnel, and competent authorities to measure service recovery and erroneous authorizations. In this way, it would be possible to assess the extent to which information overload compounds the difficulties caused by the failure, provided that confirmation channels are established before the crisis and the comparison takes into account the conditions under which decision-makers are called upon to act.
The geopolitical significance lies in the possibility that trust in instructions could become a limiting factor precisely when the state and operators must exercise their authority—including due to the influence of various actors lacking a common command structure. The attribution of individual courses of action should therefore be accompanied by communication consistent with the established facts, allowing people to recognize who is leading the response and preventing conflicting instructions from compromising its effectiveness.
The independence of recovery tools must be verified within one year
Between mid-February and mid-November 2027, the second scenario concerns the dependencies introduced by modernization, should identity, support, and registries be consolidated more rapidly than the recovery mechanisms are separated. A single outage could then affect both routine operations and the verification of the alternative, especially if digital agents acquire extensive delegated authority without leaving a reliable audit trail of the changes; integration remains, however, capable of improving security, depending on the architecture adopted. The conditional prediction is that a migration, when it also centralizes recovery tools, increases dependence on the new platform—a circumstance that can be verified by repeating the same test before and after the change and simulating its unavailability. Resources that are ready but inaccessible would reinforce this hypothesis, while maintaining timelines and control would weaken it—provided that other conditions are documented and alternative explanations are considered, such as staff shortages, incompatible spare parts, or physical problems. Shortcomings in the cost and schedule planning for the U.S. air traffic control modernization, identified by the Government Accountability Office in September 2026, highlight the need to manage the transition, though they do not prove the cause of the November outage. For Italy, any significant migration should include verification of a return to a controlled state, so that any shortcomings in the new structure can be addressed through predefined accountability and interruption criteria.
Within the Italian manufacturing sector, the availability of maintenance technicians shared by multiple companies warrants a similar review, since specialized expertise can become insufficient when many clients demand it simultaneously, even without a direct impact on the supplier. Strengthening national support and replacement capabilities—accessible to smaller companies—would help contain the spread of downtime along supply chains, protecting contracts and jobs while keeping confidential the details that would expose vulnerabilities.
Sovereignty requires joint testing and recognizable responsibilities
The link between cybersecurity preparedness and the protection of national functions can draw on the tools already provided for by the NIS2 transposition—whose Article 24 covers continuity, crisis management, and security of supply—and by the regulations on the resilience of critical entities, which address risks and interdependencies. The proposal is to coordinate exercises within their respective spheres so that the requirements of individual operators support compatible decisions when multiple services come under pressure. An initial program should focus on services where disruption and recovery directly affect people, verifying with authorities and providers what minimum capacity can be maintained, by whom, and based on what evidence. Since separately tailored plans may require the same specialists, the exercise should simulate the loss of a shared resource and communication with users and families, measuring the actual shortcomings and response times without disclosing sensitive operational details. AI could contribute through adversarial simulations and analytical support, provided that the results remain verifiable even by parties independent of the model provider. The British DROMOS project, which has experimented with AI-based review of government code while maintaining human validation, offers a model for limited-scope programs. Italy should develop forensic expertise and evaluation criteria that can be shared with its allies, while retaining the ability to recognize the limitations of the tools adopted.
The American incidents and European findings thus call for a commitment to national preparedness, the value of which does not depend on evidence of a common hostile agenda. Preserving the ability to suspend an activity, assist people, and authorize a verifiable resumption means defending the decision-maker’s autonomy when ordinary resources fail, thereby reducing the scope within which an interruption could turn into pressure on the country’s decisions. It is through this capability—which must be exercised before an emergency arises—that digital sovereignty takes on substance in the protection of the national community.