Confidential computing should be part of security planning for businesses deploying artificial intelligence because it protects sensitive information while it is being processed, not only where it is stored or how it travels. The topic was raised at a panel during the Safaricom Cybersecurity Summit 2026 in Nairobi, where Dr Markus Leberecht, a principal engineer at Intel, warned that keeping AI systems on company premises does not remove insider threats or guarantee that sensitive information is adequately protected.
The security gap between storage and processing
Traditional data-protection measures concentrate on two states: encryption at rest for stored information and encryption in transit for data moving between systems. When an application needs to compute with that information, however, it must use the data in a form the computation can act on, and that processing stage can create a security gap. This risk is especially relevant for workloads that handle confidential business records, personal information or sensitive models. AI systems increase complexity because they may process large collections of data and connect to multiple services; organisations therefore need to know which data is used, which components can access it and what happens when workloads run on infrastructure managed by another party.
How confidential computing works
Confidential computing uses hardware-based security mechanisms to create protected environments for processing data, commonly called trusted execution environments. Depending on platform and implementation, these environments can help restrict access to data and code while a workload runs, including from some privileged software layers. The protection is intended to complement—not replace—encryption at rest and in transit, identity management, least-privilege access, monitoring and sound application design. Its actual effectiveness depends on the hardware, software stack, configuration and threat model. For organisations using AI, confidential computing can be relevant when sensitive workloads run in shared or externally managed infrastructure or when limiting access by components outside a protected environment is required; it is a technical option to evaluate against specific risks rather than a universal requirement for every AI project.
Why on-premises AI is not a complete answer
Keeping models and data on premises gives an organisation more direct control over infrastructure and storage location, but it does not by itself prevent an authorised insider from misusing access, stop compromised software from acting maliciously or ensure every connected application follows the organisation’s rules. The same principle applies to cloud deployment: cloud services may offer security controls and managed infrastructure, but customers must still understand their responsibilities, configure access appropriately and review how data is handled. Data classification is an important starting point: organisations should identify which records are sensitive, determine which workloads may process them and set rules for who or what can access the information. Security and technology teams should begin with a defined use case, identify the assets to protect, the parties or software components they do not fully trust, performance requirements and the consequences of a security failure, and examine how protected workloads are deployed, monitored, updated and recovered. As businesses bring more sensitive information into AI workflows, protection of data during processing deserves a place in security planning; confidential computing offers one tool for that task, but its value depends on how it is implemented and combined with the rest of the organisation’s controls.
