Cloud failures and the growing risks of AI-generated errors expose a critical weakness in digital transformation: governments and businesses need fully operational manual systems capable of working independently alongside advanced technologies.
Microsoft’s global cloud-service disruption on November 12–13 has reinforced a warning for governments and businesses: as digital infrastructure and artificial intelligence become central to daily operations, institutions must preserve complete, independently functioning manual systems capable of sustaining essential services when technology fails or becomes unreliable.
The outage disrupted Microsoft 365 services, including Outlook, Exchange Online and Teams, affecting businesses, government agencies and public services across multiple countries. Court proceedings and healthcare administration were among the activities interrupted, demonstrating how failures in shared infrastructure can quickly become operational crises.
Microsoft attributed the disruption to a defect introduced during an earlier network-capacity expansion and exposed by a subsequent infrastructure change. The resulting server failures placed additional pressure on remaining systems. The company ruled out a cyberattack and AI agents as the cause.
Google has experienced comparable disruptions. In May 2026, failures in regional coordination infrastructure interrupted seven Workspace services for approximately 75 minutes. A September capacity test also triggered a shorter Google Drive outage. Although technically different, these incidents exposed the consequences of concentrating multiple services within interconnected digital infrastructure.
Yet infrastructure breakdown is only one dimension of the emerging risk. The expansion of artificial intelligence introduces a second vulnerability: systems may remain fully operational while producing inaccurate, fabricated or misleading information. AI hallucinations can generate nonexistent references, misinterpret records or produce convincing but incorrect recommendations. In critical workflows, such errors could influence financial approvals, healthcare administration, legal decisions or public services without any visible technical outage.
Unlike a network failure, which may trigger immediate alarms, unreliable AI output can pass unnoticed, particularly when automated systems exchange information or make decisions without sufficient independent verification.
These developments challenge the assumption that digital transformation should progressively eliminate manual processes. Removing human expertise, paper-based procedures and locally accessible operational records may improve short-term efficiency while weakening an institution’s ability to function independently.
A more resilient model would establish two complementary operational pillars: advanced digital and AI-driven systems for speed, scale and analysis, alongside fully documented, staffed and regularly tested manual systems capable of independently executing essential functions.
Such arrangements must extend beyond conventional data backups. A government agency needs procedures to process essential public requests without online portals. A bank needs tightly controlled contingency arrangements for critical transactions and approvals. Hospitals require secure downtime procedures for patient care and records, while businesses need alternative means to accept orders, authorise commitments, manage logistics and communicate with customers.
Manual operations cannot replace every digital function at equivalent capacity, particularly in real-time financial networks and complex infrastructure. They can, however, be designed to sustain defined critical services at a reduced but safe operating level, with independent authorisation, recordkeeping, human verification and subsequent reconciliation.
The two pillars should develop in harmony. Digital improvements must be accompanied by updated manual procedures, employee training, local records and periodic simulations. AI-generated decisions affecting critical activities should remain subject to appropriate human accountability and independently verifiable evidence.
This would preserve the benefits of technological progress without allowing institutional competence and operational independence to disappear behind automation.
Microsoft’s outage and Google’s earlier disruptions are reminders, not arguments against innovation. As AI advances, the greater danger may be building institutions that cannot function—or distinguish reliable information from fabricated output—without the technologies they depend upon.
The next stage of digital transformation should therefore be measured not only by how much work technology can perform, but also by how effectively governments and businesses can continue operating when it cannot be trusted or used.
Related News:
OpenAI Launches GPT-5.6 Globally as Frontier AI Race Enters a New Strategic Phase
Microsoft Commits $2bn to Extend Gulf Data-Centre Push Beyond UAE
Read Also:
AI-Led Automation Demands Stronger Cyber Resilience
AI Language Models Could Improve the Reliability of Economic Policy Analysis, ECES Study Suggests
