War risk exclusions after Nord Stream: drones and Baltic property

In March 2026 a drone that had entered Estonian airspace from the direction of Russia struck the chimney of the Auvere power station in Ida-Viru County.¹ Damage was limited and operations were not materially affected, but the incident triggered national alerts and an emergency press conference. On 23 October 2026 the prosecutor's office closed its investigation because the origin of the drone and the exact circumstances could not be established; the information gathered gave no grounds to believe that the incident was an attack directed against Estonia (Prosecutor's Office release of 23.10.2026). Similar incidents have occurred across the Baltic states, including repeated closures of Vilnius airport. Four months later, on 6 July 2026, the English Commercial Court held that Nord Stream AG's claim against its property insurers (reported at around EUR 580 million) was excluded by a war exclusion. Together, these events have moved war risk from a theoretical clause to a live coverage question for property and business interruption programmes in the region.

Why war is excluded

Many property and BI wordings contain a war exclusion, and many also deal with terrorism, sabotage or political violence, but not in a uniform way. Whether a particular wording contains a war exclusion at all, and how it defines war, has to be checked in that wording. Some exclude terrorism outright, some write it back in or offer it as a separate extension, and sabotage may be excluded, covered or not mentioned at all; definitions differ from wording to wording and from line to line. The starting point is therefore always the specific wording and any extension, not a market assumption. The rationale is actuarial: war is a systemic, correlated risk that can exceed the capacity of the private market. The difficulty in 2026 is that the line between warfare, sabotage, terrorism and criminal activity has blurred. A stray drone from a conflict between neighbouring states that hits a factory in a non-belligerent NATO state does not fit neatly into any traditional category.

The Nord Stream judgment

In Nord Stream AG v Lloyd's Insurance Company SA & Anor [2026] EWHC 1685 (Comm), decided by Dame Clare Moulder DBE sitting as a judge of the High Court, the pipeline operator claimed under its offshore operating all risks policies for damage caused by the September 2022 explosions.² The exclusion removed cover for loss "directly or indirectly occasioned by, happening through, or in consequence of war". In summary, the court held that:

• the words establish a causal test looser than proximate cause, and it overstates the position to require war to be an "effective cause" (paras 219, 243);
• it suffices that war is a "significant" cause in the sense of being noticeable or specifically accountable as a contributing factor, without any ranking of causes (paras 242, 246–247);
• there is no legal requirement that the insured property was placed in some kind of "special danger" by the war (para 295);
• the court did not need to decide which of the candidate perpetrators identified by the experts (Russia, the United States, or the Ukrainian state or a Ukrainian sub-state actor) was responsible, because for each of them the sabotage would have had the required causal connection with the war (paras 248, 261–262);
• the fact that the sabotage took place outside the territories of Russia and Ukraine did not carry particular weight (para 303);
• the damage was therefore excluded (para 647).

The judgment is a first-instance decision on a London market wording governed by English law, and its reasoning on causation is closely tied to that wording and to the expert evidence. When this post was last reviewed (4 November 2026), no appellate decision in the case had been located; whether an appeal has been brought should be checked before the judgment is relied on.

For insurers and their insureds in the Baltics, the lesson is that broadly worded exclusions may capture losses far from any battlefield where the event is connected to the wider conflict, and that uncertainty over attribution may not help the insured. For underwriters, it is equally a reminder that the result turned on specific causal language, which should be used deliberately and consistently across a programme.

How Estonian law approaches war exclusions

An Estonian court is not bound by English authority and will construe the policy under Estonian law where it governs. Several principles of the Law of Obligations Act (võlaõigusseadus, VÕS) are relevant:

Exclusion versus obligation. A war exclusion is a genuine risk exclusion: it defines the insured peril objectively and does not depend on the policyholder's conduct. The Supreme Court requires courts first to classify a clause as an exclusion or as a policyholder obligation, because the semi-mandatory limits on release for breach (VÕS § 452(2)) apply only to the latter.³ War exclusions will normally be treated as exclusions.

Burden of proof. The insurer must prove the facts on which a ground for refusal is based.⁴ In a stray drone case, that includes the facts linking the loss to war; how far the causal connection must go depends on the wording.

Contra proferentem. Under VÕS § 39(1), a standard term is interpreted as a reasonable person in the position of the other party would understand it, and in case of doubt against the party that supplied it.⁵ Where London market wording is used without adaptation, ambiguities about the scope of "war", "military operation" or "war between major powers" are likely to be resolved in the policyholder's favour, but that does not prevent prolonged litigation.

Large risks. For transport classes and for commercial insureds above the VÕS § 427(3) thresholds, the semi-mandatory provisions do not apply. War exclusions are, however, unaffected by § 427 because they are not policyholder obligations in the first place.

Local war risk extensions

In the author's experience, war risk extensions to home and commercial property policies began to appear in the Estonian market during 2026. The observations in this section are the author's general assessment of such extensions; they are not a market survey and do not describe any particular insurer's terms, and each wording has to be read on its own. The extensions are a welcome development, but they are first-generation products, and the limitations found in them typically include: cover restricted to accidental impact of a missile, drone or projectile on property that was not the intended target; extensive exclusions (civil war, terrorism, sabotage, cyber, CBRN weapons, war between major powers); exclusions of utility interruption; sub-limits below the main sums insured; and limited duration with insurer rights to amend or terminate.

The "war between major powers" carve-out raises particular interpretive issues in a NATO state. If an allied fighter shoots down a stray drone over Estonia and debris causes damage, is that a covered "military operation" or an excluded war between named powers? Does the activation of Article 5 of the North Atlantic Treaty trigger the exclusion? Current wordings provide no clear answer.

The international market

Stand-alone political violence, terrorism and sabotage cover, including resulting business interruption, is offered in the international market and can go considerably further than local extensions; what is actually available for a given risk depends on the territory, the type of asset, the sums involved and market conditions at the time of placement. War between major powers and full-scale war losses may be excluded even from such covers, so the exclusions of the specific policy have to be checked. Whether any state scheme would compensate such losses is a separate question; no general state guarantee should be assumed. Many of the scenarios actually materialising in the region (a single drone impact, an act of sabotage, an incident bordering on cyber warfare) are the kind of risk that specialist covers are designed for, and the placement should be tested against them.

Practical checklist

1. Review the war exclusion wording in property and BI policies, especially "directly or indirectly" language.
2. Compare definitions of war, terrorism, sabotage and political violence across the programme.
3. Clarify how attribution disputes will be handled and who bears the burden of proof.
4. Prioritise cover for assets near the eastern border or critical infrastructure.
5. Check whether BI and utility interruption are included in any war extension.
6. Address cyber separately: check whether the war extension excludes cyber events and how the cyber policy itself treats war.

Conclusion

The Nord Stream judgment and the Baltic drone incidents show that geopolitical risks do not respect borders and that broadly drafted war exclusions can leave policyholders without cover for events that seem far removed from the battlefield. For insurers writing Baltic risks, clear drafting is the best defence against costly disputes; for insureds, the time to review war risk cover is before the next incident.

Author: Olavi-Jüri Luik. Law as of November 2026.

Related pages

• Estonian insurance contract law
• Policy wording review under Estonian law
• Cyber insurance and NIS2 in Estonia
• Supreme Court case law
• Services for insurers

References