Regulating for Impact: Why MGAs Take A Leading Role in the Next Phase of AI‑Driven Insurance

By Andy How, Director of Insurance EMEA, Earnix

Artificial intelligence (AI) is quickly finding itself at the core of strategic decision‑making for insurance leaders. Whether it’s real‑time pricing, dynamic underwriting, or personalised customer journeys, AI is reshaping how insurance products are developed and distributed, and how insurance is being made more relevant and delivered more efficiently.

For Managing General Agents (MGAs), this shift presents an opportunity. Lean, agile, entrepreneurial and specialised, MGAs are ideally positioned to act as early adopters of AI technologies, and as catalysts for embedding innovation across the insurance value chain. But with opportunity comes responsibility, especially as regulators in the UK and EU sharpen their focus on AI’s risks and governance.

MGAs sit at the intersection of innovation and compliance. As such, their ability to navigate evolving AI regulations will be critical for their own growth, and also for their value as trusted partners to insurers and capacity providers, and for trust in the AI solutions they adopt and scale.

AI Regulation: Divergence and Direction

In the European Union, the AI Act is now in force. It implements a risk‑based framework that labels some insurance uses as high-risk – specifically, AI used for risk assessment and pricing for natural persons in life and health insurance.

The EU is also now issuing voluntary early guidance for “systemic‑risk” AI models, those with significant economic or societal footprints, urging traceability, explainability, and risk mitigation before full compliance obligations apply. This development underscores that MGAs employing decision engines for underwriting or pricing should have these controls embedded today, not tomorrow.

The UK remains regulator-led under existing laws (FCA, PRA, ICO, CMA) and the Consumer Duty. A government AI bill has been delayed into the next session; a Lords private member’s bill was reintroduced on 4 Mar 2025 but lacks government backing

This postponement comes amid some debate that the UK’s lack of binding AI regulation may leave businesses and consumers exposed. To my mind, MGAs should interpret the delay not as regulatory absence but as a signal to proceed with compliance rigor proactively and set their own standards for best practice, standardisation and vendor assurance.

Quality Data Is the Difference Between AI That Works, and AI That Hurts

In both the UK and EU frameworks, one principle remains non‑negotiable: AI’s value depends on its data. MGAs with clean, standardised, and auditable data pipelines enjoy strategic advantage: faster product iteration, more accurate risk profiling, and assurance for capacity providers. Without that data maturity, even advanced AI systems risk producing unfair, opaque, or biased outputs, exposing firms to regulatory scrutiny and reputational fallout.

Agentic and Multi‑Agent AI

The next frontier in this fast-evolving landscape is agentic AI: systems capable of autonomous decision‑making and dynamic adaptation, under expert human-in-the-loop oversight. Soon this will expand into multi‑agent ecosystems, where agents across pricing, claims, fraud, and servicing collaborate in real time.

For MGAs and insurers, the benefits are compelling: portfolio optimisation, live pricing adjustments, and hyper-personalised customer journeys. But these ecosystems amplify risk: model drift, unethical bias, opaque decisioning, coordination failures.

MGAs must embed compliance by design, transparent pipelines, ethical model training, audit trails, explainability, and documented oversight frameworks, from day one. Autonomy does not warrant opacity.

From Distribution to Decision‑Making: The Strategic MGA Opportunity

AI is not a bolt‑on, it’s a structural transformation. With rich data and disciplined governance, MGAs can evolve from price distributors to strategic co‑creators of insurance products. However, while AI is accelerating, regulation – especially in the UK, is still finding its way. For MGAs, the path forward is not to wait, but to build innovation with accountability.

That means ensuring data pipelines are standardised and audit-ready; decision models are explainable and traceable; and robust human oversight frameworks are in place. It also requires mapping compliance across jurisdictions, particularly as UK policy is still developing and potentially diverging from the EU’s AI Act, and embedding governance controls for both third-party solutions and generative AI tools. Most importantly, data ethics and fairness testing must be built into the development cycle from the outset.

Best Practice: Partnering with MGAs 

We are already working hand-in-hand with MGAs to help them seize these opportunities responsibly. By combining advanced analytics, explainable AI decision engines, and robust governance frameworks, we enable MGAs to innovate at pace while meeting the highest standards of regulatory and ethical oversight. Whether it’s building transparent pricing models, creating audit-ready data pipelines, or embedding controls into multi-agent AI ecosystems, our focus is on empowering MGAs to lead this next phase of insurance transformation with confidence that they are building best practice into their systems and processes at scale.