Sencor AI Convergence Mark SENCOR AI

Who it's for

For five kinds of organisation: a problem they have with AI output today, in the words of an outside source; rules and dates that bear on them in several jurisdictions, in the issuers' own words; and what a verdict record is designed to give them under the standard.

Sectors are named as examples where the source speaks to them.

Insurers

The problem today

Insurers also face challenges in verifying the safety measures implemented by developers for Gen AI, especially in open-source models that can be modified or misused by third parties. Given the dynamic development of the Gen AI risk landscape, insurers may also struggle to verify how businesses manage those risks.

Source: The Geneva Association, Gen AI Risks for Businesses: Exploring the role for insurance, October 2025.

Example sector: financial services and insurance.

Rules and dates that bear on them

What a verdict record is designed to give them

In the standard, a verdict record names the evidence an AI output rested on, the rule it was checked against and who issued it.

It is designed so that an independent party you authorise can rerun the check offline, taking no input from the issuer, so a claim about an AI output can be checked against the evidence it rested on rather than taken on the word of the party that made it.

Against those rules, a verdict record is designed to give an insurer evidence it can have rechecked by an independent party it authorises: what an insured's AI output rested on and the rule it was checked against, rather than the insured's own account.

Regulators and supervisors

The problem today

…monitoring efforts are still at an early stage. The report highlights challenges faced by authorities, including data gaps and lack of standardised taxonomies…

Source: Financial Stability Board, FSB outlines next steps for authorities on AI monitoring, October 2025.

Effective regulation relies on regulators having the right skills, tools, and expertise. While some regulators have been able to put the right expertise in place to address AI, others are less prepared.

Source: UK Department for Science, Innovation and Technology, A pro-innovation approach to AI regulation: government response, February 2024.

Overall, the rapid and widespread adoption of AI implies that there is an urgent need for central banks to raise their game. To address the new challenges, central banks need to upgrade their capabilities both as informed observers of the effects of technological advancements as well as users of the technology itself.

Source: Bank for International Settlements, Annual Economic Report 2024, Chapter III: Artificial intelligence and the economy: implications for central banks, June 2024.

Example sector: financial services and insurance.

Rules and dates that bear on them

What a verdict record is designed to give them

In the standard, two checkers that hold the same inputs must reach the same result and name the same step, and no step reads the checker's clock.

By design, a supervisor could have a firm's records of AI output rechecked by an independent party it authorises, from the records and the evidence they cite, without relying on the firm.

Against those rules, a verdict record is designed to give a supervisor records it can have rechecked offline by an independent party it authorises, rather than relying on a firm's own report.

Auditors and assurance firms

The problem today

Assurance providers often struggle to access the information they need to conduct effective assurance of AI systems.

Source: UK Department for Science, Innovation and Technology, Trusted third-party AI assurance roadmap, September 2025.

Rules and dates that bear on them

What a verdict record is designed to give them

By design, a verdict record carries a digest of each evidence item it cites, so an assurance provider you authorise can confirm that the evidence it is shown is the evidence the output rested on.

In the standard, a check shows where a claim came from and whether its evidence holds; it never shows that the claim is true.

Against those rules, a verdict record is designed to give an assurance provider a fixed record to examine: by design, a change made to it afterwards is detectable.

Companies building or deploying AI

The problem today

Among companies, a gap persists between recognizing RAI risks and taking meaningful action.

Source: Stanford Institute for Human-Centered Artificial Intelligence, AI Index Report 2025.

Rules and dates that bear on them

What a verdict record is designed to give them

By design, a receipt is sealed with each output, recording the evidence it rested on, the rule it was checked against and who issued it.

The standard is designed so that the check is deterministic: rerun later on the same inputs, it reaches the same result.

Against those rules, a verdict record is designed to give a provider or deployer a record of each AI output, the evidence it rested on and the rule applied, that a party it authorises can recheck later.

Enterprise buyers

The problem today

Local government is sometimes not equipped to assess the claims that are made by vendors, meaning they cannot prepare future mitigations that might be necessary for the tools they procure.

Source: Ada Lovelace Institute, Spending wisely: Redesigning the landscape for the procurement of AI in local government, November 2024.

Example sector: public sector.

Rules and dates that bear on them

What a verdict record is designed to give them

The standard is designed so that no step of a check takes its inputs from the issuer.

A buyer could therefore have a vendor's records of AI output rechecked by an independent party it authorises, without relying on the vendor's own account, by design.

Against those rules, a verdict record is designed to give a buyer a way to have a vendor's claims about its AI's output checked against the evidence by an independent party it authorises.