AI questions are landing on professional indemnity proposal forms this renewal cycle. Ed Molyneux explains why the real test of trustworthy AI won’t come from a software demo, but from your underwriter, and why auditable AI is the insurable kind.
Last week I argued that you cannot trust a large language model to ‘know’ conveyancing, and that trustworthy AI depends not on a cleverer model but on trusted data and lawyer-verified judgement the machine can reason over.
There is a commercial sequel to that argument, and it is arriving faster than most firms expect. The clearest test of whether your AI is trustworthy this year will not come from a software demo. It will come from your professional indemnity renewal.
AI questions are now appearing on PII proposal forms. Brokers and law firms tracking the market, Howden and Kennedys among them, describe underwriters asking how AI is applied across practice areas, what governance policies sit around it, and whether staff are trained to use it. Short-form declarations are becoming common, with a full proposal form typically every three years.
Double meaning
In the space of a single renewal cycle the market has moved from near-silence on AI to something more deliberate: affirmative cover that comes with conditions, sweeping exclusions where use looks uncontrolled, or silent cover that quietly depends on the firm having governance in place.
The question underwriters are circling is no longer whether your firm uses AI. It is whether you can show that use is accurate, governed and auditable.
It is worth noticing that PII carries two meanings that meet exactly here. For a conveyancer, PII is professional indemnity insurance, mandatory and claims-sensitive. In the technology world, PII is personally identifiable information, the client data that data-protection law exists to guard. They converge in the place most firms are not yet looking: ungoverned ‘shadow AI’, where a busy fee-earner pastes client details into a consumer chatbot and relies on a confident, unchecked answer.
That single action is a personal-data exposure and a professional-indemnity exposure at once, and it is precisely the kind of unmanaged risk that has always driven claims. The COLP and COFA question and the underwriter’s question have become the same question.
Governed AI is the opposite of that. It is bounded, recorded and explainable, so a firm can show what a tool did, on what basis, and stand behind the result. And that, it turns out, is an insurance property as much as a compliance one.
Unseen risk
An underwriter can only price a risk it can see. Opaque, black-box AI is close to impossible to underwrite, which is exactly why the direct-to-consumer legal tools regulators worry about carry no meaningful redress at all. AI whose reasoning and source data are traceable can be priced, and over time it is that auditability, rather than cleverness, that should bear down on premiums rather than inflate them.
The market is already showing what this looks like in practice. Orbital Witness backs its residential product with a First Title reliance policy that meets the £2 million cover the CLC requires, for under £10 a transaction, without the firm having to claim on its own professional indemnity.
That is redress engineered into the product rather than bolted on afterwards, and it is only possible because the underlying AI is reliable and transparent enough to insure in the first place. It is a glimpse of where reliable, well-governed conveyancing AI is heading.
Always-on enforcement
A necessary word of caution, because this is an area where it is easy to over-claim. Governance lowers risk; it does not abolish it, and no honest technologist should imply that a tool guarantees insurability or an automatic cut in premiums.
An open, well-designed skill that a fee-earner can ignore is a coach, not a control. Turning good practice into something an underwriter can actually rely on requires always-on enforcement and an audit trail that records what happened, which is a managed-runtime problem rather than a checklist.
The firms that will benefit are the ones that treat governed, traceable AI as the starting specification, not a badge added at the end.
This is the direction we are building toward at Moverly: open, inspectable foundations, AI that reasons over data it can prove against rules a lawyer has signed, with its rationale visible rather than hidden.
The reliability that makes an output defensible to a client is the same reliability that makes it underwriteable, and that convergence is not a coincidence. Speed only helps if the work can be stood behind.
Liability, in the end, is the prize, and the firms that can show their AI is accurate, governed and auditable are the ones both regulators and underwriters will reward. More on where that leads soon.
About the author
Ed Molyneux is co-founder and CTO of Moverly, the property intelligence platform working with LMS and Connells Group to bring structured, verified data to property transactions. He is the architect of the Property Data Trust Framework (PDTF), the open standard for machine-readable property data now being adopted across the industry. Ed writes about AI, property data infrastructure, and the future of conveyancing.
