PwC Japan is developing a system intended to continuously assess the controls over data used by AI, describing its proposed role as a “referee” for information exchanged between companies, systems and AI agents.
In an announcement on 18 November, the firm said it had obtained a patent for the underlying technology and was seeking companies to participate in developing and validating the approach.
According to PwC, Japanese patent No. 7866235 was registered on 19 May 2026 and is held by PwC Business Assurance LLC. The November announcement therefore concerns technology patented earlier this year.
What the proposed system would do
PwC describes three functions: assessing the design of data controls before use, evaluating whether those controls continue to work during use, and presenting the reasons and evidence behind assessments so people can explain decisions.
The proposed checks would examine evidence concerning data origins, consent, accuracy and completeness. Where rules are breached, PwC envisages connecting the assessment to controls that could stop data flows or use. Where rules have not been established, PwC says the system would assess risks and refer them for human judgement.
Potential applications include stopping the use of incorrect sensor data in factories or logistics operations, attaching information about source-data reliability to AI-generated management figures, and preventing AI from using personal data beyond the scope of consent.
Companies or industries would establish the rules against which the system evaluates controls.
A patent, with development still under way
The published patent describes a control system that mediates data flows between applications or services. It combines a component that implements controls with another that evaluates data and stores assessment results. Its claims also address monitoring and verification logs.
PwC’s announcement does not report independent performance results or name deployed client implementations. It says the firm is seeking partners to work on rules, implementation and validation ahead of providing the proposed capability.
Context: PwC’s physical AI push
The announcement follows PwC Consulting Japan’s July launch of a programme to train approximately 1,000 physical AI specialists, covering machines such as robots and autonomous vehicles.
The proposed referee addresses a related implementation challenge: evaluating controls over the information those systems use. PwC’s factory and logistics examples illustrate how data problems could require intervention in physical operations.
Trying to keep up with technology at Deloitte, PwC, EY and KPMG? Start here →
https://www.big4news.com/t/technology-and-ai
About the author: Claudine Cassar is the founder and editor of Big4News, covering audit, consulting, regulation and governance across Deloitte, PwC, EY and KPMG.



