Can ChatGPT Pass the SQE1? The Short Answer
No — at least not reliably, and not yet. When researchers ran ChatGPT against the SRA's official SQE1 sample questions, it scored around 50% (45 out of 90) — narrowly below the level needed to pass. Given the SQE1 pass mark typically sits around 57-62%, a score in the low 50s is a fail.
That single number is the headline, but it is not the interesting part. The interesting part is why a tool that can draft a contract clause in seconds struggles with a multiple-choice law exam — because understanding that gap tells you exactly how to use AI in your revision, and where it will quietly cost you marks if you trust it.
Why the SQE1 Is Harder for AI Than It Looks
The SQE1 is 360 single-best-answer multiple choice questions across FLK1 and FLK2. Each question is a realistic factual scenario with five options — and crucially, often more than one option is legally defensible. Your job is to pick the single best answer.
That format is brutal for large language models for three reasons:
- It tests application, not recall. AI is excellent at reproducing what the law is. SQE1 questions ask what the law does to a specific set of novel facts. The reasoning step between the two is where models slip.
- The distractors are engineered to be plausible. A model that is "90% right" still picks the wrong option when the wrong option is 85% right. Being approximately correct is worth nothing here.
- Models hallucinate black-letter detail. Ask for the threshold for a special resolution, the registration deadline for a charge, or a residual-balance limit under the Accounts Rules, and a model will sometimes invent a confident, wrong number. On a knowledge-dense exam, confident-and-wrong is the worst failure mode.
And there is a more basic point: the SQE1 is a closed-book exam sat under supervision at a Pearson VUE test centre. Even if AI could ace it, you cannot bring it into the room. The only thing that sits the exam is your trained memory.
What AI Is Genuinely Good for in Revision
None of this means AI is useless — used as a tutor rather than an examiner, it is a real accelerant:
- Explaining a concept in plain English. Stuck on the difference between a fixed and floating charge, or why a resulting trust arises? A good prompt gets you a clear, patient explanation faster than re-reading a textbook.
- Rephrasing your own notes. Paste a dense paragraph and ask for a one-line summary or an analogy. This is genuine value — summarising is what these models do best.
- Interrogating a topic. "Ask me five questions to check I understand vicarious liability" turns passive reading into active recall.
- Drafting structure. It can sketch a study schedule or reading order that you then refine.
The rule for all of these: verify anything load-bearing against an authoritative source. Treat AI output as a smart study partner who is sometimes wrong, never as the source of truth.
Where AI Will Quietly Cost You Marks
The dangerous uses are the ones that feel productive:
- Generating practice questions. This is the big trap. AI-written MCQs routinely get the "best answer" logic wrong, build distractors that are actually correct, and cite law that does not exist. Practising on them trains the wrong instincts. Practise on real exam-standard questions with written explanations instead.
- Memorising AI-summarised law. Summaries drop the exceptions and thresholds the exam loves to test. The detail you lose is exactly the detail that distinguishes the best answer from a good one.
- Solicitors Accounts calculations. Models are unreliable on multi-step arithmetic with rules attached. Accounts is a subject to learn from worked examples, not generated ones.
The Workflow That Works
Use AI for the learning phase and real questions for the training phase:
- Learn a topic from a specification-mapped source, with AI on hand to explain anything that does not click.
- Drill it on real, exam-style questions — the SRA's official samples and a proper question bank — reviewing every wrong answer.
- Test under timed conditions with full mock exams.
AI sits in step one. It does not belong in steps two and three, and that is the whole point.
The Bottom Line
Can ChatGPT pass the SQE1? Not reliably — it scores around the low 50s on official samples, below the pass mark, because the exam rewards applied judgement and precise recall under pressure, not fluent summary. Use AI as a tutor to understand the law faster, then build the skill the exam actually measures by practising on real SQE1-style questions.
That is exactly what our app and question platform are for: thousands of exam-standard MCQs with written explanations, full FLK1 and FLK2 mocks, and per-topic analytics — the trained retrieval no chatbot can give you. Start free with the 10-question readiness quiz.
FAQ
Can I use ChatGPT during the SQE1 exam?
No. SQE1 is a closed-book, supervised exam at a Pearson VUE test centre. No notes, phones or AI tools are permitted.
Will AI be able to pass the SQE1 in the future?
Models keep improving on legal knowledge, so a passing score on sample questions is plausible. But the exam is built to test application to novel facts, and you still cannot use AI in the room — so it changes nothing about how you should prepare.
Is it safe to learn the law from ChatGPT?
Only with verification. It is reliable for explaining concepts and summarising your notes, and unreliable for precise sections, thresholds, dates and case names — which it will sometimes invent. Cross-check anything you plan to memorise.