An Independent AI Evaluation Clearinghouse (IAIEC) is proposed to provide accreditation, funding, and independence in AI auditing, modeled on regulation and oversight used in safety-critical industries.

The plan prioritizes preventing conflicts of interest, encouraging competition among auditors to avoid industry capture, and broadening audits to cover bias, consumer manipulation, and privacy in addition to frontier safety risks.

The overall tone calls for a structured, government-enabled, neutral, and transparent AI auditing ecosystem aimed at boosting credibility, reducing regulatory capture, and supporting responsible AI deployment.

IAIEC would act as an auditor-of-auditors, oversee accreditation, publish redacted audit results, and ensure transparent, non-discretionary auditor assignment through mechanisms like lottery or rotation.

Historical precedents cited include the SEC's PCAOB and FINRA analogies, the FDA’s user-fee model (PDUFA), and IIHS as testing precedents, balancing government oversight with independent industry participation.

Four main recommendations: establish a federal oversight and funding framework for CAISI and IAIEC; implement formal accreditation with tiered certifications for AI auditors; secure startup and sustainable funding via scaled user fees; and use IAIEC as a neutral intermediary to assign audits non-discretionarily to accredited auditors to maximize independence.

The core problem identified is that AI safety claims are often self-certified or rely on potentially biased third-party auditors chosen by developers, unlike independent verification in automotive, toy, aviation, and pharmaceutical sectors.

A funding model is proposed where developers pay access fees into a pooled IAIEC-managed fund; auditors are paid from this pool, with participation possibly voluntary or mandated by future regulation, and audits would provide accountable, independent evaluation rather than guaranteed safety.