The NTT DATA Technology Foresight 2026 report places this question at the center of its third macrotrend: intelligence we trust. This is not just a technical issue it is a strategic business challenge. How can organizations ensure the intelligence they deploy is worthy of trust?
Twenty years ago, cybersecurity was largely focused on defending the network edge. Today, it plays a much broader role: enabling innovation, protecting reputation and helping ensure business continuity. Artificial intelligence is now moving through a similar evolution only at a much faster pace.
The conversation around AI has already moved beyond capability. It still matters whether a model works, whether it is accurate enough and which use cases it can support. But the debate has become broader. As AI systems become more autonomous, more interconnected and more deeply embedded in critical operations, the defining questions are whether they are explainable, verifiable, governable and auditable.
This is the new frontier of cybersecurity. It is about protecting intelligence itself: securing training and runtime data, protecting models from tampering, reducing the risk of inaccurate or biased outcomes, and keeping AI agents from taking actions they were not designed to perform.
Continuous assurance: From periodic control to real-time trust
Organizations are used to validating systems in cycles: annual audits, quarterly reviews and pre-launch testing. That approach was designed for relatively static environments. AI models are different. As real-world data changes, model behavior can shift in ways organizations may not predict.
Technology Foresight 2026 introduces the concept of continuous AI assurance: embedding explainability, drift monitoring, data lineage, automated evidence and governance controls directly into AI pipelines, rather than leaving them as external layers to be checked after the fact. Trust is no longer something organizations certify once. It becomes a continuous operating process, monitored and reinforced in real time.
For a mature organization, this translates into knowing not only that its credit model is 94% accurate, but also which data was used to train it. It can detect when model behavior begins to drift, explain to a regulator why a specific application was denied and activate mechanisms to reverse automated decisions when necessary.
The market is already responding
This urgency is not theoretical. The AI trust, risk and security management market, known as AI TRiSM, reached USD 2.34 billion in 2024 and is projected to reach USD 6.22 billion by 2029, with sustained annual growth of 21.6%.
According to the Technology Foresight 2026 outlook, 90% of companies will adopt AI agent governance frameworks over the next 18 months. By 2028, 85% of data products will include an AI Bill of Materials documenting how data was collected, edited and cleaned.
In other words, organizations are redefining how they demonstrate that their AI is safe, reliable and fit for use.
The role of governance and trust by design
One of the most common missteps organizations make is treating AI governance as a data-team issue or as an extension of regulatory compliance. Technology Foresight 2026 makes clear that trustworthy AI requires cross-functional governance, connecting security, compliance, architecture, ethics, operations and strategy.
That requires a shared language across functions that have historically operated in silos. A CISO needs to understand what it means for a model to have “low explainability.” A CDO must understand the governance implications of using third-party data to train an agent. A CEO needs to be able to tell the board which critical processes are automated and under what conditions they may fail.
The most mature organizations will build trust in from the start: explainable models by design, data that is traceable from ingestion, automated controls embedded in pipelines, adversarial testing before deployment and continuous authentication at runtime.
Trustworthy AI accelerates decision-making because teams are more willing to delegate. It enables personalization with confidence because data is traceable. It helps reduce fraud because behavior is observable. It is also essential to faster innovation because mechanisms exist to reverse errors when needed.
In the age of mass intelligence, trust is no longer just a value statement. It is critical infrastructure. Organizations that build it early and build it well will have the confidence to move faster when accuracy, accountability and control matter most.