Strictly necessary cookies keep you signed in and secure — those always run. Analytics cookies help us see which pages and calls to action work, and only run if you accept them. Cookie policy

    AFA Standard · In development

    Provably safe AI financial advice

    The AFA Standard scores your AI against 749 regulation-backed scenarios covering MiFID, IDD and ethics, then stays on as a runtime gate between audits.

    Have an account? Open the certification console

    749
    regulation-backed test scenarios
    MiFID · IDD · Ethics
    coverage
    One published standard
    you cannot tune to pass

    The problem

    Nobody can currently prove an AI advisor is compliant

    No sector-specific standard exists yet

    Generic AI safety frameworks don’t test for MiFID suitability, IDD disclosure, or financial advice ethics.

    Firms can’t evidence compliance to a regulator

    Without a test against a published standard and a citation trail, "our AI is fine" isn’t something a regulator or insurer can act on.

    Testing on your own is easy to get wrong

    A benchmark you write yourself, and can quietly learn to pass, is not evidence. It needs to be tested against a standard you do not control.

    How it works

    How it works

    1

    Your AI advisor is tested against the AFA matrix

    749 scenarios drawn from actual MiFID, IDD and ethics text, each with citations and pass/fail criteria, not a generic red-team pass.

    2

    Tested against a published standard, with evidence

    Every answer is judged against the rule it was tested on. You get a report you can actually show a regulator, a partner, or an insurer.

    3

    The same standard runs live, in your request path

    One API call scores an answer against the same matrix before it reaches a client — under a second, returning the rule it was checked against. It advises; your team decides whether to pass, flag, or regenerate. Testing proves the system passed. The runtime gate keeps it passing.

    What you get

    What you get

    A citation-backed report

    Evidence you can bring to a regulator, insurer, or enterprise buyer, not a self-issued badge.

    Built with financial-advice expertise

    The test matrix is grounded in real regulation and reviewed with financial-advice professionals, not written in a vacuum.

    One standard, two layers

    Periodic testing gives you the citation-backed report. The runtime gate prevents the bad answer in the first place. The same regulatory matrix sits under both.

    A related question

    Do you know where your people already send company data to an AI tool?

    Shadow AI is the tools your team already uses that were never approved: pasting a client file into a chat assistant nobody vetted. We do not have a product for this yet, we want to know if it is worth building one for your team.

    Deploying AI in financial advice? Get ahead of the standard.

    AFA Standard is in active development with financial-sector partners — tell us what you’re building.

    or