for Independent AI Assurance
Issued by AIQA Global
Artificial intelligence is now embedded in decisions that affect capital allocation, healthcare, employment, public safety, and the administration of justice. The pace of deployment exceeds the pace at which prescriptive regulation can credibly be written. In the interval, markets need a way to distinguish responsibly governed AI from the rest. Independent assurance is that mechanism. Independent assurance complements regulation; it does not replace lawful oversight.
The six principles below define what independent AI assurance is, and the conditions under which it earns the trust of the boards, insurers, investors, regulators, and procurement officers who increasingly depend on it.
Assurance is provided by a party with no financial, operational, or governance interest in the AI system being assessed.
Independence is the foundation on which every other principle rests. An assurance provider that builds, owns, sells, or operates the system it evaluates is not providing assurance — it is providing self-attestation under a different name. Independence must be structural, disclosed, and verifiable.
Assessments are based on defined, evidence-supported factors rather than narrative claims.
Responsible AI cannot be assured by adjective. The factors that determine an assessment must be specified in advance, applied consistently across organizations, and grounded in verifiable evidence. Where judgment is required, the basis for that judgment must be recorded and reviewable.
Methodology, evidence, and conclusions are documented in a form capable of independent review.
An assurance framework that cannot itself be examined offers limited assurance. Methodology must be transparent in its logic, even where specific scoring weights or proprietary algorithms remain confidential. Evidence and conclusions must be retained in a form that supports independent review, regulatory inquiry, and after-the-fact reconstruction.
Results are expressed in standardized form that permits meaningful comparison across organizations and over time.
Assurance that produces only bespoke narratives cannot inform underwriting, procurement, or capital allocation. Standardized expression — whether through scoring, certification tiers, or standardized disclosures — is what allows assurance to function as market infrastructure rather than as private consultation.
The framework evolves as AI capabilities, deployment contexts, and risks evolve.
AI is not a static technology, and assurance frameworks built for one generation of capability will not serve the next. Continuous updating must be governed by published cadence, methodology revision history, and independent technical input. Stability of result must be balanced against fidelity to the technology being assessed.
Assurance providers stand behind their methodology, governance, and representations.
Independent assurance must itself be subject to governance. Methodology changes should be documented, conflicts disclosed, and appeals or review mechanisms available where materially consequential determinations are challenged. Assurance providers should be accountable for the integrity of their process, even where they do not guarantee outcomes — assurance attests to governance quality, not to the absence of all future failure.
These principles apply to any provider of independent AI assurance, including AIQA. Independent assurance will become essential market infrastructure for responsible artificial intelligence. These principles are offered as a foundation for that work.
The institutional inheritance behind the Chicago Principles for Independent AI Assurance.
When AIQA published the Chicago Principles for Independent AI Assurance, we named them deliberately. Frameworks of this kind are usually named for what they describe — the AI Assurance Principles, the Independent Rating Standards, something abstract. We chose instead to name them for a place. Chicago is where AIQA is headquartered. It is where our first commercial client, the Am Law 200 firm Vedder, is also headquartered. And it is the city where the institutional model the Principles describe was built more than a century ago.
The history matters. The Chicago Principles do not invent a new approach to assurance. They articulate, for AI governance, an institutional discipline that Chicago institutions developed for earlier technologies and earlier markets. Three of those institutions in particular shaped the Principles, and a fourth — a more recent one — demonstrates that the Chicago tradition has continued into the modern capital markets era.
In 1893, the World’s Columbian Exposition opened in Chicago. The Palace of Electricity — lit by approximately 100,000 Edison light bulbs — was the technological centerpiece of the fair, but it was also a serious fire risk. Fire insurance underwriters sent a 25-year-old electrical engineer named William Henry Merrill, Jr. to investigate. Electricity was a new and largely uninsured technology. The risk was real, the regulatory infrastructure was nonexistent, and the people best positioned to evaluate the technology — the manufacturers themselves — had a clear interest in declaring it safe.
Merrill’s response was to found an independent testing organization. The Underwriters’ Electrical Bureau opened in Chicago in March 1894 and conducted its first test — a sample of asbestos-insulated wiring, examined for fire resistance — within weeks. By 1901, the Bureau had been chartered in Illinois as Underwriters Laboratories, Inc. The 1901 charter authorized UL “to maintain laboratories for the testing of appliances and to enter into contracts with the owners and manufacturers of such appliances respecting the recommendation thereof to insurance organizations.”
Read that charter language carefully. Independent testing of new technology. Contracted with the manufacturers, not paid by them through any consulting or implementation work. Recommended to insurance organizations. UL did not regulate. It did not certify on its own authority. It produced an independent assessment that insurers, and over time procurement officers, retailers, and regulators, could rely on. The UL Mark — a circled “UL” — today appears on more than twenty billion products annually and is recognized in dozens of countries through UL’s laboratory and certification network. Not because UL has any government enforcement power, but because the institutional model proved trustworthy enough to become market infrastructure.
Independent testing of new technology, contracted with the manufacturers, recommended to insurance organizations.
The parallels to AI governance are unusually direct. AI is an emerging technology that has outpaced its institutional infrastructure. The parties best positioned to evaluate AI governance — AI developers and deployers themselves — have a clear interest in declaring it adequate. Insurers are beginning to underwrite AI risk. Regulators are establishing frameworks but not enforcement mechanisms with teeth. AIQA seeks to build for enterprise AI governance the kind of trusted independent assurance institution that UL built for electrical safety: a private, independent, evidence-based source of assurance recognized by markets because the institution behind it has earned that recognition through transparent methodology.
Five of the six Chicago Principles trace directly to UL’s founding model. UL is independent of the products it tests. UL standards are evidence-based and reproducible. UL Marks are recognized across product categories on a comparable basis. UL standards are continuously revised through documented standards-development processes. UL stands behind its Marks through documented certification programs and federal recognition as a Nationally Recognized Testing Laboratory under the U.S. Occupational Safety and Health Administration. Independent, Measurable, Auditable, Comparable, Continuously Updated, Accountable. Five of six. The sixth, Comparable scoring across organizations and time, comes from a different Chicago institution.
In 1848, eighty-two Chicago merchants founded what would become the world’s oldest commodity exchange. The Chicago Board of Trade was created to solve a structural problem in mid-nineteenth-century American agriculture: farmers had no way to know what their crops would be worth at harvest, buyers had no way to verify what they were purchasing, and the lack of trustworthy quality measurement made commodity contracts essentially unenforceable across distance and time.
The CBOT’s solution was to establish standardized grading. By 1856, the Board had designated three quality categories of wheat with specified criteria for each. By 1858, standardized terms had been developed for forward contracts. In 1859, the Illinois legislature granted the CBOT a charter that gave the Board authority to set quality controls and to appoint grain inspectors whose decisions were binding on members. Independent grading, codified by legislative authority, with binding force. That combination — not the futures contracts themselves — was what made the modern commodity market possible. A bushel of CBOT-graded number-two wheat was the same in Chicago, in St. Louis, and in Liverpool. It was tradeable across distance and time because the grading discipline made it so.
The CBOT later developed the world’s first standardized exchange-traded futures contracts (1864), the first independent clearing organization for its contracts (the Board of Trade Clearing Corporation, established in 1925), and gave rise to the first exchange-traded options exchange (the Chicago Board Options Exchange, founded in 1973 as a CBOT spinoff). Each was an institutional innovation that extended trustworthy measurement to a new asset class. None of those innovations would have been possible without the underlying grading and inspection discipline established in the Board’s first decade.
That is the institutional heritage behind the Chicago Principle of Comparability. AI governance assurance, like agricultural commodity grading before it, must be produced on a standardized scale that allows meaningful comparison across organizations, sectors, and time. The AIQ™ Score — a 0-to-200 scale across 250 evidence-based factors — is the methodological expression of that principle. A Score of 145 means the same thing whether the organization being assessed is a bank in Chicago, a manufacturer in Stuttgart, or a hospital network in Tokyo. Comparability is the precondition for the assurance signal to function at all.
A bushel of CBOT-graded number-two wheat was the same in Chicago, in St. Louis, and in Liverpool. Comparability is the precondition for the assurance signal to function at all.
The Chicago institutional tradition did not end with UL and the CBOT. In April 1984, a 27-year-old University of Chicago MBA named Joe Mansueto founded Morningstar in his Chicago apartment with seventy thousand dollars in personal savings. The company’s founding premise was that mutual fund investors lacked access to independent, comparative information about the funds they were buying — the same structural problem of unverifiable quality that the CBOT had solved for grain a century and a half earlier, applied now to a new financial instrument that had outpaced its measurement infrastructure.
Morningstar’s response was the five-star rating system, introduced in 1985, which scored mutual funds on a standardized comparative scale based on risk-adjusted long-term performance. The methodology was published, the inputs were observable, and the resulting ratings were comparable across funds and across time. Morningstar did not regulate the mutual fund industry. It did not certify funds on its own authority. It produced an independent rating that investors, advisors, and over time fund companies themselves came to rely on. The Morningstar five-star rating is now an institutional reference standard for the U.S. mutual fund industry, despite Morningstar having no government enforcement power. The institutional model proved trustworthy enough to become market infrastructure — the same pattern UL had established for electrical safety nine decades earlier.
Morningstar matters to the Chicago Principles because it demonstrates that the Chicago institutional tradition is not a nineteenth-century artifact. The same discipline — independent measurement, transparent methodology, comparable across institutions, continuously revised, accountable through documented governance — was applied successfully to a modern capital markets category within living memory. The doctrine works. It has worked, in this city, for three different categories of unmeasurable value, across more than 175 years. The Chicago Principles propose a fourth application: enterprise AI governance.
The fourth reason the Principles are named for Chicago is harder to put into a single sentence than the first three, but it is the reason that matters most for what AIQA is trying to do.
Chicago institutional culture has historically advanced standards through demonstration before declaration. UL did not begin by lobbying state legislatures to mandate independent electrical testing. It tested, marked, and let insurers and manufacturers decide whether the Mark was worth the cost of submission. The CBOT did not begin by petitioning Congress for federal commodity-grading standards. It graded, posted prices, and let buyers and sellers decide whether the resulting market was worth participating in. Morningstar did not begin by demanding that the Securities and Exchange Commission require independent fund ratings. It rated, published, and let advisors and investors decide whether the ratings were worth consulting. In each case, the institution earned its authority by demonstrating its discipline first — then accepting, over decades, the recognition that came with that demonstration.
AIQA was founded with the same posture. We did not begin by asking regulators to mandate independent AI governance assurance. We published a methodology, scored a willing first client — Vedder, an Am Law 200 law firm headquartered in Chicago, the first major professional services firm to adopt the AIQ™ Score — and made the methodology available for any other assurance provider to adopt, adapt, or compete with. The Chicago Principles extend that posture to the doctrine of independent assurance itself. They are offered as a public reference standard, with no proprietary claim, because the strongest argument for the Principles is institutional discipline that the public can verify, not regulatory mandate that the public must accept.
This is what we mean when we say the Principles are named for the city that leads by example. Chicago institutions earned their authority by being the first to do the work, not the first to demand the rules.
AI governance is at the moment in its institutional development that electrical safety occupied in 1893, that commodity trading occupied in 1848, and that mutual fund analysis occupied in 1984. The technology is real. The risks are real. The economic stakes are real. The regulatory infrastructure is partial. The institutional infrastructure for trustworthy independent assessment does not yet exist.
Chicago institutions built that infrastructure before, three times, in this city, for earlier technologies and earlier markets. The Chicago Principles describe what it would take to build it again, for AI governance, in a way that draws on the institutional discipline those earlier institutions established.
The institutional discipline these Chicago institutions established was not parochial. UL standards are recognized across dozens of countries through international laboratory networks. CBOT-graded commodities trade in markets around the world. Morningstar operates in major financial capitals on five continents. The Chicago Principles are offered in the same spirit: a framework grounded in Chicago institutional tradition but designed for the international soft-law governance regimes in which independent assurance has become the practical bridge between voluntary commitments and credible verification.
The Principles are not original. The doctrine they articulate is more than a century old. What is new is the application: independent assurance, evidence-based and continuously updated, recognized by markets because it has earned the recognition to be relied upon.
That is why the Principles are named for Chicago. The name is not a brand. It is an institutional inheritance the Principles are intended to extend.
Issued by AIQA Global · April 2026 · Version 1.0
These principles are issued under AIQA’s name as a foundational statement for the field of independent AI assurance. They are not proprietary. Comments and engagement are welcome at principles@aiqaglobal.com.
DisclaimerAIQA Global, LLC is not affiliated with, endorsed by, sponsored by, or otherwise associated with Underwriters Laboratories Inc., UL LLC, UL Solutions Inc., UL Research Institutes, the Chicago Board of Trade, CME Group Inc., the Chicago Board Options Exchange, Cboe Global Markets, Inc., Morningstar, Inc., or any of their respective parents, subsidiaries, affiliates, or successor entities. References to these institutions in connection with the Chicago Principles for Independent AI Assurance are historical and illustrative only, made for editorial purposes to describe the institutional tradition the Principles draw upon. All trademarks and service marks referenced herein are the property of their respective owners.