The State of AI Answer Accuracy in Malaysian Insurance — A Preliminary Barometer

Picture someone in Malaysia who has been paying into a few life policies for years, suddenly reading that their insurer is in trouble. Before calling anyone, they ask a public AI assistant a simple question: if the insurer fails, how much of what they've built is actually protected? The answer comes back fast and confident, framed as if the rule were plain and settled. The fact it's describing is a public one — set out on the protection scheme's own page. The trouble is that confidence and correctness are not the same thing, and at a moment like this the gap between them is the whole question.

Public AI doesn't get Malaysian insurance wrong across the board — on the everyday details of dealing with an insurer, several tested answers tracked those details correctly. But a recurring few misstate the rules in the places a consumer acts on them: what's protected if an insurer fails, on what terms that protection applies, and when an official rule took effect. These aren't rounding errors. This is one of several sector readings in our barometer, and the misstatements here echo the shape we see across the others.

How we checked AI answers about Malaysian insurance

In June 2026 we put high-intent questions — the kind a person types before relying on protection, planning around what survives an insurer's failure, or working out what a recent rule says — to a set of public AI systems, among them ChatGPT, Gemini, Copilot, Perplexity, Grok and Google AI Overview. Each answer about a Malaysian insurance scheme was compared against that scheme's own official page: PIDM and its Takaful and Insurance Benefits Protection System (TIPS). Where an answer conflicted with the published record, we flagged it for review, and we hand-verified each finding against the scheme's official page before standing behind it.

This reading focuses on Malaysia's public protection scheme — what's protected if an insurer fails, and on what terms — with more of the sector to follow as the barometer builds month on month. It is a preliminary reading, and we'd rather say so plainly than dress it up. It covers a single month, so it's a baseline rather than a trend. And we report patterns, not a scoreboard — we name the AI systems only as the set we tested, never to rank one against another, because the more useful question is what they get wrong in common. The schemes themselves are public authorities and programmes, so we name them and cite their official pages directly; individual insurers stay unnamed by design, since the misses here are about the schemes, not any one company. The finding-level evidence and result identifiers behind each example are retained internally for audit and right-to-reply.

Where AI answers about Malaysian insurance drift

A few things recurred often enough to be worth a risk owner's attention. Each is the quieter kind of failure — an official limit, timing or date stated as something it isn't, in a place where a consumer reads it, plans around it, and only later finds the record was different. Across the reviewed examples these were specific, system-level misses, and they clustered on the scheme facts rather than the everyday details of dealing with an insurer.

On what's protected if an insurer fails, one system multiplied the real cap threefold — false reassurance at the worst possible moment. The clearest miss sat on how the protection limit aggregates. Malaysia's Takaful and Insurance Benefits Protection System (TIPS), run by PIDM (Perbadanan Insurans Deposit Malaysia), protects eligible policy benefits up to RM500,000 if your insurer fails. Crucially, those benefits aggregate — they share a single RM500,000 limit — wherever they have the same insurer member, the same risk event, the same life insured, and the same policy owner. So three death-benefit policies bought from one insurer on one life share one RM500,000 limit in total, not RM500,000 each. Asked about exactly that — three life policies with one insurer, all covering the same person — one tested system answered that the RM500,000 limit applies per policy — each covered up to RM500,000 independently. A consumer with three such policies would read that and believe up to RM1.5 million is protected when the real cap is RM500,000 total. The consequence is planning around far more protection than exists, at exactly the moment it matters most — when an insurer is failing and the family is counting on what survives.

On the headline life-insurance protection limit, one system quoted the deposit-insurance figure instead — halving the protection on offer. The second miss sat on the protected amount itself. Under TIPS, life-insurance death benefits are protected up to RM500,000. In an answer explaining PIDM/TIPS protection limits, one tested system listed the life/death-benefit limit as RM250,000 per policyholder — but RM250,000 is the figure for bank deposit insurance, a separate PIDM scheme, not the death-benefit limit under TIPS. (PIDM keeps category-specific limits across four protection funds, so the right number depends on what's being protected; here the tested life/death-benefit limit was simply misstated as the deposit-insurance figure.) The effect is the mirror image of the aggregation miss: rather than overstating protection, this answer understated it by half — telling a policyholder RM250,000 is protected when the death-benefit limit is RM500,000, the kind of understatement that could make adequate cover look like a shortfall.

On when a recent PIDM rule took effect, one system was a decade off. The third and smallest miss sat on an effective date. The PIDM Guidelines on Provision of Information on Takaful and Insurance Benefits Protection came into effect on 1 June 2024. One system stated 1 July 2014 — a decade off. The direct consumer stakes here are lower than the other two; we include it because it's a striking illustration of the same underlying habit — a confident, specific answer on a recent, official fact that turns out to be wrong by ten years. On something this checkable, the certainty is the tell.

It's worth saying what went right alongside these, because public AI is not uniformly wrong here. On the everyday details of dealing with an individual insurer — how long a complaint decision should take, how policies are issued, the cooling-off or free-look window — several tested answers tracked those everyday insurer details correctly. They correctly cited the headline RM500,000 TIPS limit and the exclusion of investment-linked units from it; the new complaint-handling rule taking effect in April 2026 (a decision within 5 working days for simple cases, up to 20 for complex ones); the roughly 15-day cooling-off period; and the 60-day timing for escalating an unanswered complaint to the Financial Markets Ombudsman Service (FMOS) — each of these stated correctly. The misses didn't spread evenly across everything we asked; they concentrated on PIDM's TIPS protection facts — what's protected, and on what terms — not on the day-to-day insurer details.

Why AI answer accuracy is a governance risk for insurers

None of these answers were written by the scheme, and none can be edited by it. That's precisely the difficulty. PIDM can keep its RM500,000 TIPS limit, its aggregation rule, and its guideline dates plainly stated on its own pages — and a consumer can still arrive having been told that RM1.5 million is protected, or that the life-insurance limit is only RM250,000, by a system the scheme has no contract with and no visibility into. The records were right. The representation circulating about them wasn't. For insurers, the scheme administrator, and the financial institutions whose customers sit downstream of the scheme, the consequences — a misdirected expectation about how much is protected, an avoidable complaint, a customer who feels misinformed about their cover — are theirs to understand, evidence, and manage, regardless of who typed the answer.

Most governance frameworks weren't built to see this. They scrutinise what the institution publishes. This risk lives one step outside that perimeter, in the growing share of consumers who ask an AI before they ask the insurer or the scheme.

The Malaysian context makes the issue sharper: this scheme touches high-stakes, emotionally loaded decisions — what survives an insurer failure — where a consumer is least able to sense that a confident answer is off. The errors we saw aren't random noise; they cluster on what's protected and on the terms that protection comes with, the area where a misinformed consumer has a concrete grievance. A RM500,000 aggregate cap stated as RM500,000 per policy is the kind of thing a family plans around. So is a RM500,000 death-benefit limit quoted as RM250,000 — adequate cover made to look like a shortfall, or the reverse. The same shape shows up in our Singapore insurance reading, which suggests this isn't a quirk of one market but a pattern in how public AI handles precise, official facts.

How to govern AI answers about your institution

The answer isn't to chase the AI, which can't be corrected the way an institution's own website copy can, but to govern the surface with the same seriousness applied to any channel that speaks in the institution's name. In practice that's a repeatable loop: watch what the major systems are telling consumers about your products and the schemes they sit alongside; check it against the official published facts; separate the genuine misstatements from the false alarms before acting on either; work out why a real one drifted, since a stale source and outside misinformation call for different fixes; rank what could actually mislead a consumer — a protection cap stated too high, or the headline limit halved — ahead of what's merely imprecise; and keep a dated record of what was said and when. Done steadily, an open-ended worry becomes a managed process — the institution finds out before the customer does, corrects the sources it controls, and can show its work to a board, or to a regulator, when asked what it's doing about AI.

Read on

This is an early read from an ongoing barometer Lawnise is building on how public AI answers questions about Malaysia's insurance sector, with more categories to follow over time. If you'd like to see what public AI is currently saying about your own institution and the schemes around it — checked, the way these were, against the official published facts — we can scope a private baseline: sector-context, no obligation.