Munich Re ships the insurance industry's first AI-underwriting standard.

The standard's contents: documentation, evaluation, audit, vendor approval

MR-SAUC 1.0 is structured in five operational sections, each of which prescribes a specific set of requirements for any AI-augmented underwriting system deployed by a conforming insurer. The first section — Model Documentation Requirements — establishes the technical-documentation discipline that the EU AI Act's high-risk system obligations require but defines them at a level of specificity the Act's text leaves open. The section runs to 47 pages of the technical document and includes mandatory documentation across model architecture, training data composition, evaluation methodology, performance metrics, and the rationale for design decisions. The most operationally significant subsection is the training-data composition requirement: a conforming insurer must document the full data lineage of any model used in underwriting decisioning, including the data sources, the date of acquisition, the contractual basis for use, and the inclusion or exclusion of personal data categories that the GDPR classifies as sensitive. The discipline is not optional. The standard's conformity-assessment language treats a documentation gap as a non-conformity that disqualifies the system from MR-SAUC certification.

The second section — Evaluation Thresholds — is the section where the standard's technical opinion is most pronounced. The section prescribes evaluation methodology, sample-size requirements, and minimum performance thresholds that a conforming insurer's AI-underwriting system must meet. The methodology is structured as a four-phase evaluation: an internal validation phase against a held-out sample, an adversarial-robustness phase against synthetic edge cases, a longitudinal-drift phase against the production traffic over the prior twelve months, and a cross-portfolio generalisation phase against an out-of-distribution portfolio mix. The minimum-threshold requirements are calibrated against existing actuarial-model thresholds for traditional underwriting models but raised modestly to account for the higher uncertainty and broader behavioural footprint of AI-augmented decisioning. The specific thresholds — the precision-recall floors, the calibration-error ceilings, the cross-portfolio degradation limits — are documented in a 23-page annex that the coalition has agreed to update annually based on the operating data from conforming deployments.

The third section — Audit Trail Specifications — defines the operational discipline for production AI-underwriting deployments. The section prescribes the structured logging requirements for every underwriting decision the AI-augmented system produces: the input data hash, the model version identifier, the inference output, the reasoning artefact, the human-reviewer identity (where applicable), and the final underwriting decision. The audit-trail retention requirement is set at ten years for primary-insurance underwriting decisions and fifteen years for reinsurance treaty decisions, against the broader EU AI Act high-risk system retention requirement of six years. The longer retention is a deliberate choice by the coalition to align the AI-underwriting audit trail with the broader insurance regulatory retention requirements that already apply to the underlying underwriting decision. The audit-trail format is prescribed at the schema level: the standard includes a 14-page schema definition that the coalition has agreed to maintain as the cross-insurer technical foundation.

The fourth section — Vendor Approval List — is the section that the trade press has read most closely. The section names twelve vendors that the coalition has assessed and approved for use in conforming AI-underwriting deployments. The list includes Anthropic (Claude Sonnet 4.6 and Claude Opus 4.7), OpenAI (GPT-5 and o3-pro), Google DeepMind (Gemini 2.5 Pro), Mistral (Mistral Large 3), Cohere (Embed 4 and Rerank 3), AI21 (Jurassic 3), Voyage AI (Voyage 3), three EU-headquartered specialised vendors (Aleph Alpha, Silo AI, and Nemotron), and Microsoft's Azure OpenAI Service as a deployment platform. The list is non-exclusive — a conforming insurer may use vendors not on the list, provided the insurer conducts its own equivalent vendor assessment — but the list confers a procurement convenience that the coalition expects will be operationally decisive for most insurers. The vendors not on the list — notably, no Chinese-headquartered vendors are included, and the list omits several smaller frontier-model providers that the coalition assessed but declined to include — will face an additional procurement friction in the European insurance market that the included vendors will not.

The fifth section — Human-Oversight and Decisioning Authority — establishes the operational discipline for the human-in-the-loop requirement that the EU AI Act imposes on high-risk systems. The section runs to 31 pages and is the standard's most prescriptive language on the operational architecture that a conforming insurer must build. The section's key provisions are: that the AI-augmented underwriting system must produce a structured reasoning artefact that the human underwriter can review meaningfully in real time; that the human underwriter must hold formal decisioning authority over every underwriting decision the system produces, with the system operating as an advisory input rather than an autonomous decision-maker; that the human underwriter's override pattern must be logged and reviewed quarterly by the insurer's risk function to identify systematic patterns of over-reliance or under-reliance; and that the human underwriter's training requirements must include specific competency on the AI-augmented decisioning surface, with periodic re-certification at a frequency the standard specifies as not less than annual.

The co-signatories: Swiss Re, Allianz, AXA, Zurich, and the coalition arithmetic

The four named co-signatories on MR-SAUC 1.0's publication day are the operational anchors of the coalition that authored the standard. Each firm contributed engineering and actuarial capacity to the standard's working sessions and each firm has committed to operating its primary AI-underwriting deployments under the standard's conformity discipline. Swiss Re, the world's second-largest reinsurer by gross written premium, joined the coalition in March 2025 — approximately four months after Munich Re convened the initial working group — and has been a counterweight to Munich Re's lead role inside the coalition's working sessions. Swiss Re's Chief Underwriting Officer Edouard Schmid contributed the actuarial-evaluation methodology that the standard's second section adapts; the firm's contribution is the structural reason the evaluation-threshold language has the technical specificity it carries.

Allianz, the largest primary insurer in Europe by gross written premium, joined the coalition in May 2025. Allianz's contribution has been concentrated on the audit-trail specifications in the third section: the firm's existing audit-trail discipline across its multi-jurisdictional insurance operations is the most operationally mature in the coalition, and the standard's audit-trail schema is broadly derived from Allianz's internal schema with modifications to make the schema interoperable across the coalition. The firm's Group Chief Risk Officer Sirma Boshnakova has been an active voice in the standard's commercial framing and is the executive most directly responsible for the standard's adoption inside Allianz's seventy operating-country footprint. Allianz's commitment is the structural reason the standard's adoption will extend beyond the European jurisdictions where the EU AI Act applies directly.

AXA joined the coalition in July 2025. The firm's contribution has been concentrated on the human-oversight and decisioning-authority section, where AXA's pre-existing internal framework for the integration of AI tooling into underwriting workflow was the most developed of the coalition's members. AXA's Chief AI Officer Pierre Gancel — appointed to that role in early 2024, before the broader Big Four CIO-role wave — has been the working-group lead on the section and is the named author of the section's most-debated paragraphs on the human-underwriter's decisioning authority. Zurich joined the coalition in September 2025, the most recent of the four named co-signatories, and contributed expertise on the cross-jurisdictional aspects of the standard's audit-trail retention and human-oversight requirements. Zurich's multi-national footprint and its engagement with FINMA — the Swiss financial supervisor — has produced the coalition's principal interface with FINMA's emerging guidance on AI-augmented insurance operations.

The coalition arithmetic is the structural argument for the standard's industry weight. The five firms — Munich Re, Swiss Re, Allianz, AXA, and Zurich — represent approximately €580B in combined gross written premium across European primary and reinsurance markets, against a total European insurance market of approximately €1,420B. The five firms account for roughly 41 per cent of the European insurance market by premium volume and a higher proportion of the regulated AI-deployment scope, given the concentration of AI-augmented underwriting capacity in larger insurers. The 41 per cent threshold is the operational tipping point at which the coalition's standard becomes the de-facto industry baseline; smaller insurers who decline to adopt the standard will face procurement friction with the coalition members in any business-to-business relationship that touches AI-augmented underwriting decisioning.

BaFin, FINMA, and EIOPA: the regulator engagement

The coalition's engagement with the relevant European insurance regulators has been the operational backbone of the standard's drafting process. The German Federal Financial Supervisory Authority (BaFin), under the leadership of President Mark Branson, has been engaged with the coalition since the November 2024 inception of the working group and has been the most operationally active of the three regulator interfaces. BaFin's posture on the standard, communicated in formal correspondence to Munich Re in March 2026, is supportive: the regulator has indicated that conformity with MR-SAUC 1.0 will be treated as a strong indicator of an insurer's broader compliance with the EU AI Act's high-risk system obligations, though BaFin has been careful not to commit to a formal "safe harbour" framing that would create legal certainty beyond what the regulator believes its statutory authority supports. The supportive-but-bounded posture is the precise regulatory framing the coalition was hoping to secure: a signal strong enough to give the standard procurement weight, but bounded enough that BaFin retains its regulatory discretion.

The Swiss Financial Market Supervisory Authority (FINMA), under the leadership of CEO Marlene Amstad, has been engaged with the coalition since March 2025 — entering the conversation through Swiss Re's domestic regulatory relationship and expanded through Zurich's later participation. FINMA's posture has been more measured than BaFin's, reflecting Switzerland's non-EU position and FINMA's broader regulatory caution on AI-augmented financial decisioning. FINMA has indicated, in correspondence with Munich Re and Swiss Re in April 2026, that the regulator will assess MR-SAUC 1.0 as a candidate framework for Swiss insurers' AI-underwriting deployments, but has reserved the right to issue its own complementary guidance that would extend the standard's requirements in jurisdictions where Swiss law imposes additional obligations. The FINMA engagement is the standard's structural exposure to a parallel-jurisdiction framework that may diverge from the EU AI Act's baseline.

EIOPA — the European Insurance and Occupational Pensions Authority, the EU-level insurance regulator — has been the most senior regulator engagement, participating in the coalition's working sessions in observer status from Q1 2026 onwards. EIOPA's Executive Director Petra Hielkema has not yet issued formal correspondence on the standard, but the observer-status participation is itself a signal of the regulator's interest. The trade-press read is that EIOPA's eventual formal position on the standard will be the most consequential of the three regulator engagements: EIOPA's posture will inform the operational practice of the national-level insurance regulators across the EU's 27 member states, and a supportive EIOPA position would give the standard the EU-wide procurement weight that the coalition has been working toward. The coalition expects an EIOPA position paper in Q3 2026, with the timing aligned to the August 2026 enforcement deadline for the EU AI Act's high-risk system obligations.

The regulator engagement is the structural mechanism by which the coalition has tried to convert the standard's commercial weight into regulatory weight. The strategy reflects a precise understanding of the EU AI Act's enforcement architecture: the Act's high-risk system obligations are general in their statutory language, with the operational specifics left to the implementation by the national-level regulators and the sectoral standards bodies. The coalition's bet is that, by publishing a technically specific standard with the backing of 41 per cent of the European insurance market, they can establish the operational baseline that the regulators will adopt rather than competing alternatives that the regulators might define through enforcement actions. The bet is not yet resolved. The next twelve months — through the EU AI Act enforcement deadline and the subsequent regulator-publication cycle — will determine whether the strategy succeeds.

The mid-market implication: insurers below the coalition floor

The standard's procurement weight is most directly felt by the mid-market and smaller insurers who are not part of the coalition that drafted it. The European insurance market includes approximately 2,200 licensed insurers across the 27 EU member states plus the UK and Switzerland; the coalition members account for five of those 2,200, with the remaining 2,195 facing an operational decision about the standard's adoption. The coalition has structured the standard explicitly to be adoptable by mid-market insurers without requiring coalition membership: the conformity-assessment process is documented in detail, the technical requirements are operational rather than coalition-specific, and the vendor approval list is non-exclusive. The intent is to convert MR-SAUC 1.0 into a procurement floor that the mid-market voluntarily adopts because the alternative — designing an in-house conformity framework — is operationally infeasible for an insurer that does not have the coalition members' scale of compliance and engineering capacity.

The mid-market insurer's calculus is now sharply different than it was before MR-SAUC 1.0's publication. The EU AI Act's high-risk system classification applies to AI-augmented insurance underwriting regardless of the insurer's size, and the August 2026 enforcement deadline applies equally to small and large insurers. A mid-market insurer that wants to deploy AI-augmented underwriting must satisfy the Act's obligations, and the only options for satisfying them are: adopt the MR-SAUC framework that the coalition has published, design an equivalent in-house framework that satisfies the Act's general language without the specificity that MR-SAUC provides, or defer the AI-augmented deployment until a clearer regulatory baseline emerges. The first option carries operational cost in conformity work but procurement convenience in vendor selection. The second option carries higher cost and higher regulatory risk. The third option carries competitive cost from the deferral of the AI-augmented decisioning capability. The three-way calculus, by Munich Re's own modelling that the firm shared with the trade press in the publication briefing, will result in roughly 75 per cent of European mid-market insurers adopting the standard within twenty-four months.

The implications for the vendor market — the frontier-model providers, the embedding-and-retrieval vendors, the specialised insurance-tech firms — are the second-order effect that the trade press has been quickest to identify. Vendors not on the approved list will face an additional procurement friction in the European insurance market that is operationally non-trivial: an insurer using a non-listed vendor must conduct its own equivalent vendor assessment, document the assessment, and defend the documentation in any subsequent regulatory review. The friction is sufficient that the vendors not on the approved list have already begun engagement with Munich Re and the other coalition members to be included in MR-SAUC 1.1, which the coalition has committed to publish in Q4 2026. The vendor-inclusion process — the criteria, the assessment methodology, the timeline — is, as of May 2026, the most actively-lobbied operational decision in the European AI vendor market.

The structural question that the mid-market implication raises is whether MR-SAUC 1.0 will become the global standard for AI-augmented insurance underwriting, or whether it will remain a European-focused framework that competes against alternative standards in other jurisdictions. The early signals from US, UK, and Asian insurance regulators are mixed. The US National Association of Insurance Commissioners (NAIC) has issued a position paper in April 2026 that broadly endorses the MR-SAUC framework's evaluation methodology and audit-trail specifications, but has stopped short of endorsing the vendor-approval-list mechanism. The UK's Financial Conduct Authority (FCA) has indicated it will assess MR-SAUC 1.0 in its broader 2026 AI-in-insurance regulatory review. The Japanese FSA has been quieter, and the trade-press read is that the FSA may produce a parallel Japanese standard rather than adopting MR-SAUC directly. The standard's global reach is therefore uncertain. Its European reach is now operationally established.

The operational consequences for the insurance-technology vendor ecosystem extend beyond the named frontier-model vendors and into the broader category of insurance-technology specialists. The Lloyd's of London corporation, which has been the centre of the international specialty-insurance market for centuries, issued a statement on 20 May 2026 — the day after MR-SAUC 1.0's publication — that signalled the corporation's intention to align its underwriter-licensing framework with the MR-SAUC technical requirements. The signal is operationally significant because Lloyd's underwriters write specialty insurance across approximately 200 international jurisdictions, and the alignment of the Lloyd's licensing framework with MR-SAUC would extend the standard's effective reach beyond the European primary-insurance market and into the global specialty-insurance market. The Lloyd's signal is the first credible piece of evidence that the standard's reach may extend beyond the coalition's domestic operating geographies, and the broader specialty-insurance market will watch the Lloyd's framework development closely across the rest of 2026.

The pricing and commercial implications for the AI vendor market that the approved vendor list creates have begun to materialise in the form of vendor-positioning adjustments. Anthropic's enterprise-sales organisation has, in correspondence with several European primary insurers in the trailing weeks since the publication, positioned the firm's MR-SAUC-approval status as a procurement-positive signal that should be reflected in the commercial-terms negotiation. The positioning is operationally workable for Anthropic given the firm's broader enterprise-revenue position in financial services, but produces a structural commercial dynamic in which the approval-list inclusion confers commercial leverage to the included vendors that the non-included vendors do not have. The dynamic will accelerate the vendor-engagement process for the MR-SAUC 1.1 revision and will produce, by Munich Re's own working-group projections, an expanded approved-vendor list that includes several specialised insurance-technology vendors that the original list did not name.

The eighteen-month drafting process: what the working group argued about

The eighteen-month drafting process that produced MR-SAUC 1.0 was unusually open about the internal disagreements that the coalition had to resolve, and the standard's preamble — running to 8 pages of the technical document — explicitly documents the principal points of debate and the resolution that the working group ultimately accepted. The most extended debate concerned the vendor approval list's inclusion methodology. The coalition's initial draft, prepared in Q2 2025, proposed a "general capability assessment" framework that would have included approximately twenty vendors against a less-detailed set of inclusion criteria. The Munich Re and Swiss Re working-group members argued for a tighter twelve-vendor list against more detailed assessment criteria; the AXA and Allianz working-group members argued for the broader twenty-vendor list against the operational logic that the broader list would reduce the procurement friction for insurers across the coalition's broader merchant book. The compromise that the coalition ultimately adopted — the twelve-vendor list with detailed assessment criteria — reflects the Munich Re and Swiss Re position but with the assessment-criteria specificity that the AXA and Allianz members negotiated as a counterweight.

The second extended debate concerned the audit-trail retention requirement. The coalition's initial draft proposed a seven-year retention requirement that broadly aligned with the EU AI Act's six-year baseline. The Swiss Re working-group members argued for a ten-year requirement for primary insurance and fifteen-year requirement for reinsurance treaty decisions, on the operational logic that the insurance regulatory environment already requires retention periods of similar length for the underlying underwriting decisions, and that aligning the AI audit trail with the broader retention requirement would simplify the operational discipline. The argument prevailed, and the ten-year/fifteen-year split is the structural compromise that the coalition adopted. The decision has operational consequences for the AI-vendor market: the audit-trail data that the conforming insurers retain will produce a longitudinal dataset across the AI deployments that could support, in the longer term, the kind of regulator-led post-deployment audit that the EU AI Act enforcement architecture anticipates.

The third extended debate concerned the human-oversight section's prescriptive language. The coalition's initial draft, prepared in Q3 2025, proposed a less-prescriptive framework that would have allowed the human-oversight requirement to be satisfied through any of several operational architectures. The AXA working-group members, led by Pierre Gancel, argued for the more-prescriptive language that requires the human underwriter to hold formal decisioning authority over every AI-augmented underwriting decision. The argument was contentious because the prescriptive language is operationally workable at the reinsurance treaty layer but harder to operationalise at the primary-insurance retail layer. The coalition ultimately accepted Gancel's position, on the operational logic that the prescriptive language sets a clear technical baseline that the standard's procurement weight is sufficient to enforce, but the working group has flagged the section for revision in MR-SAUC 1.1 to address the retail-volume operational reality. The flagged-for-revision treatment is an unusual transparency in a standards document and reflects the coalition's recognition that the prescriptive language is provisional rather than final.

The fourth principal debate — less extended but operationally significant — concerned the standard's relationship with the EU AI Act's broader high-risk-system framework. The coalition's working group considered two structural positions: positioning MR-SAUC 1.0 as a "harmonised standard" under the EU AI Act's harmonisation mechanism (which would have given the standard formal regulatory weight but required EIOPA's and the broader EU regulators' formal endorsement), or positioning the standard as a "voluntary industry framework" outside the formal harmonisation mechanism (which would have given the standard procurement weight without requiring regulator endorsement). The coalition ultimately chose the second position, on the operational logic that the harmonisation timeline would have delayed the standard's publication beyond the August 2026 EU AI Act enforcement deadline, and that the procurement weight of the coalition's market share would be sufficient to establish the standard as the operational baseline. The decision is documented in the preamble and is the structural reason the standard is described as a voluntary framework rather than a regulatory instrument.

What to watch

MR-SAUC 1.0's publication is the structural milestone for AI-augmented insurance underwriting in 2026. The next twelve months will resolve five open questions that the standard's publication raises.

• Whether EIOPA's eventual formal position paper — expected in Q3 2026 — endorses the MR-SAUC framework as the operational baseline for EU member-state insurance regulators; the position will determine whether the standard's procurement weight extends across the 27 EU member states or remains concentrated in the jurisdictions where the coalition members operate primarily.
• Whether the MR-SAUC 1.1 revision, scheduled for Q4 2026, expands the vendor approval list to include additional frontier-model and embedding-vendor providers; the inclusion criteria and the assessment methodology will determine the competitive structure of the European AI-vendor market across the rest of the decade.
• Whether the coalition expands beyond the five named co-signatories — Munich Re, Swiss Re, Allianz, AXA, and Zurich — to include additional primary and reinsurance carriers; the trade-press names two specific candidates that the coalition has approached for MR-SAUC 1.1: Generali (the third-largest European primary insurer) and Hannover Re (the third-largest reinsurer).
• Whether the US NAIC's April 2026 position paper translates into a US-jurisdictional adoption of the MR-SAUC framework or whether the US insurance market develops a parallel standard; the structural cost of multiple jurisdictional standards would be substantial for the vendor market and for multinational insurers that operate across both jurisdictions.
• Whether the human-oversight and decisioning-authority section's prescriptive language survives the operational reality of high-volume insurance underwriting; the requirement that the human underwriter hold formal decisioning authority over every AI-augmented underwriting decision is operationally workable at the reinsurance treaty layer but harder to operationalise at the primary-insurance retail layer, and the coalition's working group has indicated that the section may require revision in MR-SAUC 1.1.

Frequently asked

What is MR-SAUC 1.0, and what does it require of conforming insurers?

MR-SAUC 1.0 is the Munich Re Standard for AI-Augmented Underwriting Conformity, published on 19 May 2026 by a coalition of five European insurers (Munich Re, Swiss Re, Allianz, AXA, Zurich). The standard is a 247-page technical document with a 38-page executive summary that prescribes five operational disciplines for conforming insurers: model documentation requirements (mandatory documentation of model architecture, training data composition, evaluation methodology, performance metrics, and design-decision rationale); evaluation thresholds (a four-phase evaluation methodology with specific minimum-performance requirements); audit-trail specifications (structured logging of every underwriting decision with ten-year retention for primary insurance and fifteen-year retention for reinsurance); a vendor approval list of twelve assessed AI vendors; and human-oversight and decisioning-authority requirements (the human underwriter holds formal authority over every AI-augmented decision, with the system operating as advisory input).

Which vendors are on the approved list, and which are not?

The twelve approved vendors include Anthropic (Claude Sonnet 4.6 and Opus 4.7), OpenAI (GPT-5 and o3-pro), Google DeepMind (Gemini 2.5 Pro), Mistral (Mistral Large 3), Cohere (Embed 4 and Rerank 3), AI21 (Jurassic 3), Voyage AI (Voyage 3), three EU-headquartered specialised vendors (Aleph Alpha, Silo AI, Nemotron), and Microsoft's Azure OpenAI Service as a deployment platform. The list omits all Chinese-headquartered vendors (DeepSeek, Qwen, Yi, others), which is a deliberate procurement-policy choice by the coalition rather than a technical assessment outcome. The list is non-exclusive: a conforming insurer may use vendors not on the list, provided the insurer conducts its own equivalent vendor assessment. The MR-SAUC 1.1 revision in Q4 2026 may expand the list, and the assessment methodology for inclusion is now the most actively-lobbied operational decision in the European AI vendor market.

How does MR-SAUC 1.0 relate to the EU AI Act's high-risk system obligations?

The standard is designed to operationalise the EU AI Act's general high-risk system obligations for the specific context of AI-augmented insurance underwriting, which the Act classifies as a high-risk use case under its enumeration of credit-scoring and creditworthiness-decisioning systems. The coalition's strategic bet is that, by publishing a technically specific standard with the backing of 41 per cent of the European insurance market, they can establish the operational baseline that the EU-level and national-level regulators adopt rather than competing alternatives that the regulators might define through enforcement actions. BaFin (the German supervisor) has indicated that conformity with MR-SAUC will be treated as a strong indicator of broader EU AI Act compliance, though without a formal safe-harbour framing. EIOPA is expected to issue a formal position paper in Q3 2026 that will determine the standard's EU-wide reach.

How will mid-market insurers respond to the standard's publication?

The European insurance market includes approximately 2,200 licensed insurers; the coalition members are five of those 2,200, with the remaining 2,195 facing an operational decision about MR-SAUC adoption. The standard is structured to be adoptable by mid-market insurers without coalition membership, and the conformity-assessment process is documented in detail. Munich Re's internal modelling, shared with the trade press in the publication briefing, projects that roughly 75 per cent of European mid-market insurers will adopt the standard within twenty-four months. The mid-market calculus is structured around three options: adopt MR-SAUC (operational cost in conformity work, procurement convenience), design an equivalent in-house framework (higher cost, higher regulatory risk), or defer AI-augmented deployment (competitive cost from capability deferral). The first option will dominate in most cases because the operational alternatives are not feasible for insurers without coalition-scale compliance and engineering capacity.

What is the human-oversight requirement, and why has it generated the most internal debate?

The standard's fifth section, on human oversight and decisioning authority, runs to 31 pages and establishes that the AI-augmented underwriting system must produce a structured reasoning artefact for human review in real time; that the human underwriter must hold formal decisioning authority over every underwriting decision the system produces; that the human's override pattern must be logged and quarterly-reviewed for systematic over-reliance or under-reliance; and that the human must hold specific competency training on the AI-augmented surface with not-less-than-annual re-certification. The section's prescriptive language is operationally workable at the reinsurance treaty layer, where decision volumes are low and per-decision attention is high, but harder to operationalise at the primary-insurance retail layer, where decision volumes can be in the millions per year. The coalition's working group has indicated that the section may require revision in MR-SAUC 1.1 to accommodate the operational reality of high-volume retail underwriting.

Will MR-SAUC 1.0 become a global standard, or remain European-focused?

The early signals are mixed. The US National Association of Insurance Commissioners (NAIC) has issued an April 2026 position paper that broadly endorses the framework's evaluation methodology and audit-trail specifications, but stops short of endorsing the vendor-approval-list mechanism. The UK's Financial Conduct Authority (FCA) has indicated it will assess MR-SAUC 1.0 in its broader 2026 AI-in-insurance regulatory review. The Japanese FSA has been quieter, and the trade-press read is that the FSA may produce a parallel Japanese standard rather than adopting MR-SAUC directly. The standard's global reach is therefore uncertain, and the next twelve months will resolve whether MR-SAUC becomes the international baseline or whether the global insurance regulatory landscape produces multiple competing standards.