Cloud outage liability: who answers when Microsoft fails
On 12 November 2026 a software fault took Exchange Online and Outlook offline for about 36 hours. A legal analysis published shortly afterwards argues that responsibility for such outages often lies with the operator rather than the provider.

What happened on 12 November 2026
A software fault disabled the Microsoft services Exchange Online and Outlook for roughly 36 hours. Emails did not arrive, and business processes stalled in many companies and public bodies. In an interview with tagesschau24, the security expert Dennis-Kenji Kipker described the effect on many firms as digital brain death.
The consequences went beyond lost working hours. According to the lawyer Christian Heinelt, writing in Legal Tribune Online, the British NHS had to activate emergency plans and postpone treatments. In Germany the transfer of medical findings between practices and laboratories came under strain. Media reports say 15 people suffered health damage at a leisure park because an email alarm failed during the outage and a hotel pool was overchlorinated.
Who carries the risk: provider or operator?
Anyone who creates or maintains a source of danger must make sure nobody is harmed. German criminal law calls this a duty of a guarantor (Garantenstellung). It applies to digital risks as well, so a pool operator must secure the IT that controls and monitors the chlorine supply. A medical practice must keep emergency plans and alternative routes for urgent findings.
Software vendors usually answer for a single component, while the operator controls the whole system. The author concludes that Microsoft staff are probably not exposed to criminal liability for such a fault. The operator's duty of care often outweighs the responsibility for a software error, especially where the operator relies on one supplier without a fallback.
The picture can change if a vendor knows of a defect and neither fixes it nor warns customers. The Federal Court of Justice (BGH) expects manufacturers to monitor their products and act on known dangers. On the civil side, the new Product Liability Act covers software from 9 December 2026. It does not apply to this outage, and it does not cover pure financial loss such as a standstill.
The role of management and IT service providers
Regulation such as the German NIS-2 implementation act assigns responsibility for IT security to management explicitly. Entities in scope must prepare for crises and manage supply chain risks, and management must approve and oversee these measures. Criminal law does not follow this rule directly, yet the author sees it as a legislative value judgement that carries weight, also for companies outside NIS-2.
IT service providers who plan, set up and maintain a system can hold a guarantor position themselves, depending on their contract. Management cannot hand over its responsibility wholesale. It must select the provider with care, monitor it and steer it.
The key question is whether those responsible could rely on a single transmission channel. Large outages are foreseeable, and Microsoft 365 already failed in January and August of this year. A critical alarm that runs only through a cloud mailbox is a single point of failure.
What to check in your company
- List which safety-critical or time-critical messages depend on one cloud service, such as alarms, lab findings or order confirmations.
- Give each of these a second, independent channel, for example a phone call, SMS or a separate system.
- Write an emergency plan for the loss of central cloud services and rehearse it with the people who would use it.
- Document who in management approved the plan and who monitors your IT providers.
- Check your contracts with service providers for the scope of what they plan, run and monitor.
Whether your own setup meets legal duties depends on the facts, so involve a lawyer for the assessment. Companies commonly review these steps after an outage of this kind. The author of the analysis advises routing safety-critical messages over at least two independent channels and keeping an emergency plan ready for the loss of central cloud services.
A useful test is to imagine that your main mailbox is unavailable for a day and a half, as it was in November. Ask which alarms stay silent, which findings cannot be sent and who would notice. The answers show where a second channel is needed first.
Using company knowledge with AI and planning for outages
AI tools that read company documents add another dependency to the list. Anyone introducing them should ask how they behave when a connected service such as Exchange is unavailable, and who approves actions that change data.
The model catalogue of souveraen.ai includes local models, models hosted in Germany and external EU models with zero data retention. The platform connects sources such as SharePoint, Exchange and file servers read-only, with existing permissions. Every answer cites its source, and agents wait for approval before writing to a system. For setups that need to run without any outside connection, the Air-Gap Enterprise option is sized by quote.
Sources
- Globale Großstörung: Wenn der IT-Ausfall zum Strafbarkeitsrisiko wird, Legal Tribune Online


