Mayflower Specialty, Ltd. and Hadron launched what they describe as the first dedicated affirmative AI liability program in the United States on June 24, 2026. The program, issued by Hadron and underwritten by Mayflower, provides explicit coverage for directors and officers (D&O), employment practices liability (EPL), and errors and omissions (E&O) for enterprise AI deployments. The move shifts AI liability transfer from ad-hoc endorsements toward standalone underwriting.
For insurance professionals, the launch of an affirmative AI liability product is a notable development in the AI for Insurance space, as it creates a clearer path for quantifying residual legal and governance risk from deployed models. The policy suite includes affirmative grants for management liability, employment practices, and professional liability, paired with a difference-in-conditions and excess layer that drops down where legacy policies are silent, sub-limited, or exclusionary.
Product structure
The program targets enterprise companies deploying AI and covers the management-liability triad. Business Wire reported that the product is backed by institutional reinsurance partners and that placement was structured by Aon Reinsurance Solutions. Hadron has institutional backing that includes more than $250 million in committed capital from Altamont Capital Partners and other investors.
"Every major technology wave has created a new class of liability, and the underwriters who showed up early with genuine standards defined that market for a generation," said Jeremy Epstein, Founder and CEO of Mayflower Specialty. Pete Buccola, Group Chief Underwriting Officer of Hadron, said the need exists to both limit exposure and consider underwriting approaches for consumers.
Underwriting and risk controls
Mayflower underwrites AI-specific risks-including model bias, model drift, and hallucinations-using an auditable scoring engine aligned to NIST and ISO standards. The carrier cites McKinsey's 2025 State of AI report that 88% of organizations have integrated AI into at least one business area as evidence of market need. The scoring engine's reliance on public standards suggests a checklistable baseline for controls that vendors and internal ML teams can target when seeking coverage.
For risk managers, the underwriting approach means that evidence-auditable logs, evaluation metrics, and governance artifacts-will likely be required to rate exposures. The product's appetite for specific failure modes maps traditional professional-liability constructs onto model-specific risks, a pattern seen in other jurisdictions that typically starts with narrow, conservatively priced capacity.
What to watch
Three indicators will determine the product's impact:
- Policy wording and claims definitions-whether terms like "hallucination" and "bias" are operationalised into measurable trigger events.
- Underwriting data and scoring criteria-whether the auditable engine publishes benchmarks or remains proprietary.
- Reinsurance capacity and placement dynamics-whether institutional reinsurers expand participation, which would broaden pricing and limits.
Why this matters for insurance professionals
The launch signals that underwriting markets are beginning to map model-specific failure modes onto traditional liability lines. For insurance professionals, the practical takeaway is procedural: documenting evaluation datasets, drift-monitoring telemetry, and model-change logs creates the artifacts that underwriters will likely require as affirmative AI products scale. If the product gains traction, it could make certain governance failures financially transferable and others explicitly uninsurable, altering the residual risk calculus for boards and CISOs.
Your membership also unlocks:
