Verification infrastructure for AI output
Beyond proof, AI needs verdicts.
Verdicts unlock AI's potential.
In the verdict record standard, a sealed receipt travels with each AI output: it records the evidence the output rested on, the rule it was checked against and who issued it. By design, an independent party you authorise can recheck a receipt later, offline, without relying on Sencor AI or the producer.
- claim
- refund approved, €120
- evidence
- sha256:9f2c…e41a
- rule
- amount-matches/v1
- result
- holds
- IETF Internet-Draftdraft-wright-verdict-record-02
- Public test caseswith a reference checker
- 13 research papersopen access; twelve also in 15 languages
- 24 UK patent applicationsfiled at the UK IPO
Why now
Why verification, and why now.
- 01
AI is moving fast.
Since the publication of the 2025 Report, general-purpose AI capabilities have continued to improve, driven by new techniques that enhance performance after initial training.
Source: International AI Safety Report, 2026 Report: Executive Summary, February 2026.
The adoption of LLMs and gen AI tools is proceeding at such breathtaking speed that it easily outpaces previous waves of technology adoption…
- 02
Its safety concerns are real.
Current AI systems sometimes exhibit failures such as fabricating information, producing flawed code, and giving misleading advice. AI agents pose heightened risks because they act autonomously, making it harder for humans to intervene before failures cause harm.
Source: International AI Safety Report, 2026 Report: Executive Summary, February 2026.
Responsible AI is not keeping pace with AI capability, with safety benchmarks lagging and incidents rising sharply.
- 03
Rules for AI are being adopted around the world.
Globally, legislative mentions of AI rose 21.3% across 75 countries since 2023, marking a ninefold increase since 2016.
Source: Stanford Institute for Human-Centered Artificial Intelligence, AI Index Report 2025, 2025.
The AI Act (Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence) is the first-ever comprehensive legal framework on AI worldwide.
Source: European Commission, AI Act, last updated 19 October 2026.
The Ministry of Science and ICT ("MSIT") announced that the National Assembly's plenary session passed the Basic Act on the Development of Artificial Intelligence and the Establishment of Foundation for Trustworthiness ("AI Basic Act") on December 26, 2024.
- 04
Compliance is becoming a requirement.
Once an AI system is on the market, authorities are in charge of market surveillance, deployers ensure human oversight and monitoring, and providers have a post-market monitoring system in place. Providers and deployers will also report serious incidents and malfunctioning.
Source: European Commission, AI Act, last updated 19 October 2026.
The MAS Guidelines apply to all financial institutions (FIs) and all forms of AI technologies. They set out MAS' supervisory expectations for FIs to manage risks arising from AI use, while allowing them to tailor their approaches according to their risk profiles, including the scale and nature of their AI use.
Consistent therewith, all Insurers authorized to do business in this state are expected to develop, implement, and maintain a written program (an "AIS Program") for the responsible use of AI Systems that make, or support decisions related to regulated insurance practices.
- 05
Money at risk needs cover, and insurers find AI risk hard to verify.
As it is difficult for insurers to verify Gen AI risks and how businesses manage them, Gen-AI-related insurance may experience serious information asymmetry. Insurers may therefore be reluctant to offer high coverage limits, as in the early days of cyber insurance.
AI introduces emerging risk dimensions that do not fit neatly within traditional insurance boundaries. These include cyber and fraud risks, liability exposures linked to algorithmic failures or bias, intellectual property disputes, and non-physical business interruption.
- 06
Underneath all of it: verification.
Each step above turns on the same question: can AI's output be trusted at the speed and volume it is produced?
While every output waits for a person to check it, checking is the bottleneck.
When a verdict can be checked by a machine, checking stops holding AI back.
That is how verification unlocks AI's utility: its throughput and its efficiency.
How it works
By design, a receipt travels with every output.
- 01
Produce
An AI system produces an output: a decision, a figure, a summary.
- 02
Seal
A receipt is sealed with it, recording the evidence, the rule applied and who issued it.
- 03
Recheck
An independent party you authorise reruns the check offline, from the record and the evidence it cites, and reaches the same result.
- 04
Reassess
When the state of a cited record changes, every record that cites it shows the change the next time it is checked, to the depth the checker follows.
Reassessment, by design
Designed so a change reaches what was built on it.
In the verdict record standard, records cite what they depend on. When the state of a cited record changes, every record that cites it shows the change the next time it is checked, to the depth the checker follows. No notice has to be sent and no list of dependants has to be kept up to date.
What a receipt does not claim
By design, it vouches for origin and evidence, not for truth.
A receipt shows how an AI knows something and whether that evidence holds. It is designed to be tamper-evident, not tamper-proof: a change is detectable and provable by an independent party you authorise.
Why neutrality
A checking layer one party can switch off will not be trusted by the rest.
Sencor AI's design has nothing to switch off, little to surrender, and rules published openly at the IETF rather than held by one company or country.
The open standard
One rulebook others can implement.
The verdict record is an individual Internet-Draft at the IETF. Its checks are published with public test cases and a reference checker, so a separate implementation can be compared against the answers the draft text gives.
Research
The Verification Intelligence series
For people who check, insure, regulate or build AI systems.
If you need records of AI output designed so that an independent party you authorise can recheck them, talk to us.
Who it's for: insurers, regulators, auditors, companies building AI, and buyers →