QuavaBETA

Trust

Last updated: 17 June 2026

You're revising for an exam where the gap between the right answer and a plausible wrong one is the law — and law that was correct last year may not be this year. So before you rely on anything here, the fair question is simple: how do we know it's right?

Most revision material now passes through AI, and much of it can't be trusted — invented authorities, confidently wrong answers, law that has quietly gone out of date. MCQs are a particular weak spot: nearly all AI models output slop when asked to write a plausible, exam-realistic MCQ.

We're not claiming to have completely solved these problems. We're showing you how we've worked to close down the scale of many of them — how we place AI in the hands of people who understand law and technology, and of experts who can properly audit its output.

We've spent hundreds of hours thinking about legal accuracy and the shortcomings of AI: how systems and prompt chains can synthesise expertise in law and code; how we base our materials in the actual law and keep AI scaling to that law; how we keep materials accurate to the date the SRA actually examines; who stands behind it; how we use AI; and what happens when we get something wrong.

The law as it stands on 13 March 2026

The SQE does not examine the law as it stands today. It examines the law of England & Wales in force on a fixed date, set by the SRA before each assessment window. For July 2026 SQE1 that date is 13 March 2026. A change that took effect the day after — however prominent — is not examinable; a change from the year before still is. Revising to the wrong date is among the easiest ways to lose marks on law you actually know.

So we write to that date, not to the present. Our standard: content states the law in force on the examinable date for your sitting, and where the law has recently changed we flag it rather than leave you to infer which version applies.

Holding to that takes active work. We track legislation, SRA updates and new case law, and before each cycle we run a currency review across our questions and lessons, checked against primary sources. We publish the result as a separate, dated record — our law-currency log — so you can see which changes we've accounted for, and as at when. If you find content that's fallen behind, the Report an issue control on the page itself is the fastest way to tell us — see How we correct mistakes.

We build from primary sources — not AI assertions

We build from primary sources, and we show them to you. Our materials don't just state the law; they carry the authority behind it, so you can verify any point yourself rather than take it on faith:

  • Statutes, cited to section. Where a rule comes from an Act, we name the Act and the section and link to the provision on legislation.gov.uk, set to the point-in-time version in force on your examinable date — not the current consolidated text, which may already have moved on.
  • Cases, cited to authority. Leading cases are given by name and neutral citation and linked to the report (BAILII or the official series), so you can check the court, the year and what was actually decided — not a paraphrase of it.
  • Practice, mapped to official SRA materials. Our questions and worked scenarios are constructed against the SRA's published SQE assessment specification and official sample and specimen questions — not invented look-alike content that drifts from how the exam is actually written.
  • Verified, not assumed. Every citation is checked against the primary source as part of review. A reference points to a real section or a real case — not a plausible-looking one. Where we can't stand a point up against a source, it doesn't ship.

Who writes and stands behind our content

Quava's content is produced and editorially led by Angus Livingstone, who holds 20 years' experience in legal education. He was elected Teaching Fellow at City Law School at 23.

Editorial responsibility is named, not anonymous: a real person's judgement sits behind what you read, and that person is accountable for it.

Nothing reaches you that isn't built on a core corpus of legal rules drawn from primary sources. This core corpus sits outside the products you see in this application; it is Quava's source of truth. It is not 100% perfect, but it is a major upgrade in legal accuracy compared with the standard output of AI models. Everything is anchored to our core corpus.

How we use AI — with a human holding the line

We use AI in producing content: drafting from primary sources, restructuring, surfacing candidate sources, and pressure-testing explanations. We're open about that because the alternative — pretending otherwise — is exactly the dishonesty this page exists to refuse.

The line we hold is bright and it does not move: a qualified human verifies the core law; AI never decides what the law is. No statement of law, citation, or answer reaches you on the strength of a model's confidence alone. AI helps us work faster and find what we'd otherwise miss; it is never the authority.

Model-assisted review — Fable Update, June 2026

Anthropic's frontier model Fable 5 was available in the UK from 9 June until 12 June. To be clear — we used Fable intensively in this period, as a check pass over all our materials. It reviewed every word of every lesson, concept and MCQ, and the result was substantial: refinements to over 2,000 MCQs — sharper wording, tighter distractors, clearer explanations — and light-to-medium redrafts of roughly 30% of our lesson content. Every change was reviewed and signed off by us. You can read the full record here — [link to the report].

The in-app Report an issue feedback control

How we correct mistakes

We will get things wrong. An accuracy rate of 99.5% (our loose, and currently unproven, estimate) still implies over 15,000 words of wrong law across the 3 million words of lessons and MCQs. Finding that 0.5%, and reducing it towards zero, is our obsession.

  • Every content page can be reported, in one tap. There's no separate form to find and no email to compose. Every content page in Quava — every lesson, every question, every worked answer and explanation — carries a built-in Report an issue control. If you can see it, you can flag it, without leaving the page.
  • We triage by stakes. A point of law that could mislead you in the exam is treated as priority over a typo. Any report of this nature is reviewed within 48 hours by a human, and corrections are made at that time.
  • We credit you. If your report leads to a correction, we credit your membership renewal with £5. You can claim as many £5 credits as you like, up to the value of your monthly subscription.

Independence

Quava is an independent revision provider. We are not affiliated with, authorised, endorsed, or approved by the SRA or Kaplan, and we don't suggest otherwise. "SQE", "SQE1" and related marks belong to their owners; we refer to them only to describe what we help you prepare for.

What we claim

What we claim: that our content states the law as in force on your examinable date, is built from and verified against primary sources, and is corrected openly when it's wrong.

What we don't claim: that Quava is legal advice, that it's a substitute for the SRA's own materials, or that using it guarantees a pass. Outcomes depend on you. Our full terms are at Terms.

Accessibility

We're working toward full WCAG 2.2 AA conformance. We haven't yet completed an independent audit against every criterion, so we won't claim we're there — but we've built for it from the start. What's in place today:

  • Dark theme throughout. A dark view across the whole app, not just selected pages.
  • Adjustable text size. Increase or decrease text to suit how you read.
  • Comprehensive keyboard navigation. Full keyboard operation and shortcuts everywhere, so the app is usable without a mouse.
  • Reduced cognitive load. Material is broken into focused slides rather than dense walls of text.

If you hit an accessibility barrier, the Report an issue control flags it straight to us — see How we correct mistakes.

Accountability

If something here doesn't hold up — the law, a source, a claim on this page — write to accountability@quava.app and it will reach management directly, not a queue.