TL;DR:

• The Lloyd’s Market Association, working with consultancy Barnett Waddingham, has published an AI Adoption Toolkit to help managing agents formalise governance as AI moves from pilots into core underwriting workflows.
• A parallel LMA survey covering more than 60% of Lloyd’s stamp capacity shows a clear majority of firms now have or are developing structured AI frameworks, with mandated human oversight of AI outputs becoming the default rather than the exception.
• The toolkit is principles-based rather than prescriptive and is organised around five themes: governance and accountability, risk tiering, data protection, training, and pragmatic adoption.

The Lloyd’s Market Association has released an AI Adoption Toolkit designed to help the insurance market’s managing agents operationalise AI governance, as a parallel survey shows Lloyd’s firms have moved decisively out of the pilot phase and into production deployment across pricing, exposure management and claims.

Where the market is on adoption

The LMA-Barnett Waddingham survey, covering over 60% of the Lloyd’s market’s stamp capacity, found that formal AI governance frameworks have become the norm rather than the exception — a meaningful shift from the scattered pilot landscape the LMA reported a year ago. Firms are now running AI in core actuarial workflows, with pricing, exposure management and claims the three most common deployment areas. Responsibility for governance typically sits jointly with senior technology leaders and dedicated committees, and firms are increasingly focused on data privacy, cyber security and third-party risk as adoption scales.

Regulators are closely watching. The UK’s Financial Conduct Authority, Bank of England’s PRA and the Lloyd’s Corporation itself have all flagged AI model risk as a priority, and the LMA’s timing — publishing formal guidance rather than high-level statements — reflects direct demand from members for implementation-ready frameworks. The survey result that most firms are mandating human oversight of AI outputs signals the market has converged on augmentation rather than full automation as the near-term deployment pattern.

What the toolkit covers

The toolkit is organised around five core themes: governance and accountability; risk tiering; data protection, security and intellectual property; training and awareness; and pragmatic adoption. It is principles-based rather than prescriptive, so syndicates at different stages of AI maturity can use it without having to adopt a single organisational model. Sanjiv Sharma, LMA’s head of actuarial and exposure management, said the emphasis was on embedding appropriate controls early and scaling adoption in a structured way.

Risk tiering is the most procedurally significant theme — the expectation is that firms classify AI use cases by potential impact on reserving, capital, risk selection or client outcomes and apply governance proportionately. That maps closely to the tiered approach the EU AI Act has adopted and the PRA’s own risk-based stance on model risk management.

Looking Forward

For UK insurers — and for the dozens of international insurers that underwrite at Lloyd’s — the toolkit is a concrete benchmark that boards, internal audit and the PRA can reference in live supervision. Expect the five themes to become de facto standards for insurance-sector AI governance beyond the Lloyd’s market, particularly for mid-market firms that lack the resources to build governance frameworks from scratch. Firms still piloting should use the toolkit structure to compress their onboarding timeline.