VISIONQUANTECH BLOG

Can AI Write Content for Law Firms? An Edinburgh Firm's Honest Answer

2026-11-18·VisionQuantech Team

TL;DR: Yes — AI can write the first draft of a law firm's blog posts, practice-area pages and FAQs, and Google does not penalise content for being AI-assisted. What it cannot do is check Scots law, cite a real case, or sign off on advertising rules: 206 AI-hallucination cases reached court filings by mid-2025, with sanctions from $2,500 to $31,100. The model that works is AI drafts plus solicitor review.

A solicitor's desk in a bright Edinburgh office, a laptop showing a clean blog-draft document with a short checklist beside it, a printed statute book and reading glasses on the desk, daylight from a tall Georgian sash window
Fiona's desk on George Street, Edinburgh — the draft, the checklist and the statute book. (Illustration; video: can AI write content for law firms, in 25 seconds.)

Meet Fiona — and the Monday a rival firm got her client

Fiona runs a four-partner family and conveyancing firm off George Street in Edinburgh. Her firm is good at the work and invisible on the internet. The blog on her website was last updated eleven months ago, with a post about stamp duty that is now wrong for Scotland — Scotland has its own Land and Buildings Transaction Tax, not stamp duty, and the post never said so.

On a Monday morning in November, a woman selling a flat in Stockbridge phoned a rival firm instead of Fiona's. When Fiona's receptionist asked how she had chosen them, the woman said she had asked ChatGPT who to call for a conveyancing solicitor in Edinburgh. ChatGPT named three firms. Fiona's was not one of them.

That is the question this piece answers: can AI write content for a law firm — Fiona's firm, in Edinburgh, under Scots law — and if so, what does it cost, what are the traps, and what does "done properly" actually look like?

The short answer: yes, for drafts — no, for the final word

An AI content writer is a large language model trained to predict the next most likely word. Given a topic, a practice area and a city, it can produce a structured 1,500-word first draft in under five minutes, with headings, a FAQ section and the long-tail keywords clients actually type — phrases like "how long does divorce take in Scotland" or "what is a Home Report in Edinburgh".

That is the useful part. Here is what it cannot do, and every honest guide agrees on this list:

- It cannot verify the law. A language model predicts likely word sequences; it does not retrieve facts. It will invent case citations that look real, quote statutes that do not exist, and state English law as if it applied in Scotland. Scotland has its own legal system, its own courts and its own regulator, the Law Society of Scotland. A draft that says "stamp duty" where the answer is Land and Buildings Transaction Tax is not a small error — it is the kind of error that tells a reader the firm does not know its own jurisdiction. - It cannot know your firm. It has no access to your cases, your results, your voice or your local knowledge. A page that could belong to any firm in any city is exactly the kind of page Google now calls commodity content. - It cannot take responsibility. Under the American Bar Association's Formal Opinion 512, issued 29 July 2024, lawyers remain responsible for AI-generated content published under their name — including marketing content. The duties are competence, confidentiality, transparency and billing. In Scotland, the same principle applies through the Law Society of Scotland's conduct rules: the solicitor signs off, the solicitor owns it.

The winning model, in every serious guide published this year, is the same: AI drafts, a solicitor reviews. AI does the first 60%; a human does the final 40% that protects trust. A firm that publishes raw AI output and calls it an SEO strategy typically sees flat or declining organic traffic within six months, as Google's quality systems catch up with undifferentiated volume.

What Google actually says in 2026 — and what the data shows

This is the part most firms get wrong, because the rumour mill runs ahead of the documents. Google's position has been consistent since February 2023, when it wrote: "Appropriate use of AI or automation is not against our guidelines." The March 2024 spam-policy update added the line that matters: generating many pages "without adding value for users" is scaled content abuse. The violation is volume without value — not AI.

On 1 October 2026 Google updated its guidance on generative AI content again, adding language from its Search Quality Rater guidelines: models predict likely word sequences rather than retrieve facts, their output can contain hallucinations, and it is, in Google's word, "critical" to manually fact-check and review all AI-generated content before publishing — down to the title tags and the alt text. Google almost never uses the word "critical" in its help documents. It used it here.

And the ranking data backs the documents, not the rumours. On 16 October 2026, CLM Sequoia scored every law firm page ranking in the top eight Google results for 2,418 competitive legal searches — 6,044,850 words in total. The average page was 27.9% AI-written. More than one in five ranking pages — 21.6% — was 76% to 100% AI-generated. And on the 528 searches tracked in both February and October, the share of mostly-AI pages was 19.7% before the year's updates and 19.8% after. Firms whose pages averaged 51% or more AI content kept their rankings through the September 2026 spam update at the same rate as firms under 10%. If Google had quietly started punishing machine-written legal pages, that number would have cratered. It moved by three points in the wrong direction for that story.

The honest reading: AI content is not penalised. Thin, generic, wrong content is — and AI makes it cheap to produce at scale, which is why the October guidance reads like a warning label. The firms ranking today are not the ones avoiding AI. They are the ones fact-checking it.

There is a second reason to care, beyond Google. A 2026 survey by iLawyerMarketing found roughly 42% of consumers now use ChatGPT to research attorneys, while Google still leads at 72%. Clients ask an AI answer engine who to call, and the answer names a few firms. Fiona's firm was not named — not because her firm is worse, but because her firm has almost nothing for the answer engines to cite.

The Edinburgh maths: what a blog post really costs a small firm

Before prices, do the maths on the status quo, because that is what Fiona was actually paying.

A senior solicitor writing a 2,000-word guide — say, "how long does divorce take in Scotland" — spends four to six hours on it: research, draft, check, rewrite. At $350 an hour, the figure US guides use, that is $1,400 to $2,100 of billable opportunity per article. At Edinburgh rates the number is lower in pounds but the shape is identical: two posts a month costs a partner roughly a full working week a year. Clio's 2024 Legal Trends Report found attorneys average only 2.9 billable hours a day as it is. A firm that asks its solicitors to write the blog is asking its most expensive people to do its cheapest job.

The alternative most firms try first is the agency retainer: $4,000 to $8,000 a month for four to eight posts, which works out at $800 to $1,000 per article. A freelance legal writer charges $200 to $500 per article in the US, £200 to £500 in the UK; a UK agency bills £500 to £2,000 per blog post. The content-mill end of the market — $50 to $150 an article — produces generic pages with minimal SEO value, which is exactly the commodity content Google's October guidance warns about.

The price ladder: what AI content for a law firm costs in November 2026

Rung 1 — DIY tools, $20 to $70 a month. ChatGPT Plus at $20 a month drafts unlimited first drafts, outlines, FAQ lists and social captions. Jasper Pro is $59 per seat a month on annual billing ($69 month-to-month) with brand-voice controls; KoalaWriter runs $7.65 to $41.65 a month and is built for SERP-driven long-form drafts. What you buy is speed. What you do not buy is legal accuracy, jurisdiction checks or review — the solicitor still does all of that, and the first draft still needs a heavy edit. Reviewers consistently report the same two complaints about Jasper: output that reads generic or repetitive, and long-form posts that lose coherence after 600 to 800 words. Its ratings tell the same split story — 4.7 out of 5 on G2 across 1,317 reviews, 4.8 on Capterra across 1,855, but 3.2 on Trustpilot across 4,144 reviews, where billing friction dominates.

Rung 2 — legal-specific AI writers, $199 to $999 a month. WebWriter.ai (formerly JurisContent) is built for law firms in all 50 US states: statutes and case law cited, bar-rule-safe language, state-specific pages. Starter is $199 a month for 10 articles; Professional is $399 for 30; Enterprise is $999 for 100. Content packs add five articles for $99. This rung is the first one that understands jurisdiction as a feature rather than an accident — but it is built for US state law, not Scots law, so an Edinburgh firm still needs its own review pass for every Scottish point.

Rung 3 — humanised done-for-you, $300 to $800 an article. Lexicon Legal Content puts the price of quality humanised AI legal content at roughly $300 to $800 per article, and warns that content priced at $50 to $100 is almost certainly surface-only — a wording polish on an unchecked draft rather than a production method with attorney knowledge, factual verification and verifiable human authorship built in. At this rung you are paying for the review, not the words.

Rung 4 — an installed AI Content Writer, $2,500 to $12,500 setup. VisionQuantech's AI Content Writer is installed for the firm rather than rented as a tool: Starter $2,500–$4,500 one-time setup with $500–$800 a month care, Growth $4,500–$6,500 with $999 a month care, All-in-One $8,500–$12,500 with $1,999 a month care. Founding clients get 20% off the first project setup — first ten clients only. The difference from the rungs above is that the system is configured to the firm's voice, practice areas and jurisdiction, and the review workflow is part of the installation. It is a service, not a login: there is no self-serve tool to buy on a card.

For Fiona's firm, the comparison that matters is simple. Two solicitor-written posts a month costs roughly a week of partner time a year. A DIY tool costs $20 a month plus that same review time. A legal-specific writer at $399 a month produces 30 articles — more than Fiona needs — but none of them checked against Scots law. The done-for-you rung is the only one where the check is somebody's job.

The comparison table

OptionWhat you payWhat you getThe catch for an Edinburgh firm
ChatGPT Plus$20/moUnlimited first drafts, outlines, FAQsNo legal accuracy, no jurisdiction check, no review
Jasper Pro$59/seat/mo annualBrand voice, templates, marketing agentsGeneric output on complex topics; coherence fades past ~800 words; G2 4.7/5 but Trustpilot 3.2/5
KoalaWriter$7.65–$41.65/moSERP-driven long-form drafts with live citationsStill a draft machine — no solicitor review, no bar-rule check
WebWriter.ai$199–$999/moLaw-specific engine, statutes cited, 10–100 articles/moBuilt for US state law, not Scots law
Lexicon-style done-for-you$300–$800/articleAttorney-reviewed, humanised, E-E-A-T built inPer-article cost; volume stays low
VQ AI Content Writer$2,500–$12,500 setup + $500–$1,999/mo careInstalled system, firm voice, review workflow includedService, not self-serve; founding 20% off first 10 only

The four traps every law firm must survive

Trap 1 — hallucinated citations. The most documented failure in legal AI. More than 206 cases of AI hallucinations in court filings were documented by mid-2025, with sanctions ranging from $2,500 to $31,100. The famous one is Mata v. Avianca in 2023, where two attorneys filed a brief citing cases ChatGPT had invented and were sanctioned. Marketing content is lower-stakes than a court filing, but the mechanism is identical: a confident, plausible, entirely fictional citation on a page carrying your firm's name. The test before signing anything: ask the vendor to show you one article where the AI cited a case, and ask who verified it.

Trap 2 — the wrong jurisdiction. This is the trap that only exists for firms outside England and Wales, and it is the one that will catch an Edinburgh firm first. Scots law is a separate legal system: different courts, different conveyancing (missives and the Home Report, not English exchange of contracts), different tax (Land and Buildings Transaction Tax, not stamp duty), different divorce procedure. A generic AI draft will default to English law every time, because most of its training data is English. The test: give the vendor the sentence "what tax do I pay when I buy a flat in Edinburgh" and read the answer. If it says stamp duty, the system is not configured for your firm.

Trap 3 — advertising rules. Every state bar regulates lawyer advertising, and the shared themes are the same everywhere: no false or misleading statements, no guaranteed outcomes, no unsubstantiated claims of specialisation. AI is trained to be persuasive, and left unchecked it will write copy that implies guaranteed results and overstates expertise. The Florida bar's 2025 ethics opinion clarified that AI-assisted content still requires a reviewing attorney; New York's 2025 formal opinion requires disclosure where content could be mistaken for the attorney's personal expertise. In Scotland, the Law Society of Scotland's conduct rules carry the same duty. The test: does the vendor's workflow include a named reviewing solicitor, or does the draft go straight to publish?

Trap 4 — thin content that ranks, then slides. LawRank's analysis found pure AI content may rank initially, then slide over a six-month window as Google's systems assess depth. Human-authored content holds rankings and engagement over time. The reason is E-E-A-T: legal content is Your Money Your Life content, and Google evaluates it with higher scrutiny than a recipe blog. A page with no author, no experience, no local fact and no original point of view is commodity content — and Google's own Danny Sullivan said in April 2026 that the shift is toward non-commodity content, meaning content a reader cannot find in equivalent form elsewhere.

The 5-step switch: moving a firm's content to AI without breaking anything

Step 1 — pick one practice area, not the whole firm. Fiona started with conveyancing, because it is the firm's highest-volume work and the questions are the most predictable: Home Reports, missives, LBTT, timelines. One area, ten questions, one month. If the system cannot answer ten conveyancing questions correctly, it cannot answer fifty questions across five areas.

Step 2 — write the firm's voice down before the AI writes a word. Two pages: how the firm talks to clients, words it never uses, claims it never makes, and the disclaimer that goes on every page ("general information, not legal advice"). A tool with no voice document will write in the voice of the average internet law firm — which is to say, no voice at all.

Step 3 — run the jurisdiction test. Give the system five questions that only a Scots-law answer can pass: LBTT on a first purchase, the Home Report, missives, the Scottish divorce process, and a no-win-no-fee personal injury claim. If any answer defaults to English law, the configuration is wrong and no amount of volume will fix it.

Step 4 — install the review loop, and name the reviewer. Every piece gets a named solicitor's eyes before publish, with a checklist: every statute checked, every case verified, every claim about the firm true, every advertising rule respected. The ABA's Opinion 512 makes this a duty, not a nicety. Fifteen minutes of review per article is the figure the better vendors quote — and it is fifteen minutes well spent, because it is the difference between a draft and a publication.

Step 5 — publish little, measure, then scale. Four posts in month one, not forty. Watch three numbers: whether the pages appear in AI answers to the questions they target, whether enquiries mention the article, and whether any page slides after six months. Firms that maintain active blogs generate 68% more qualified leads than firms that do not publish consistently — but that number belongs to firms that publish well, not firms that publish a lot. Scale what the review loop can hold, not what the tool can produce.

Frequently asked questions

Can AI write content for law firms?

Yes — AI can draft blog posts, practice-area pages, FAQs and social posts for a law firm, and Google does not penalise content for being AI-assisted. It cannot verify the law, cite real cases or take responsibility for advertising compliance. The model that works is AI drafts plus solicitor review, with every statute and citation checked before publish.

Will Google penalise my law firm's website for AI content?

No. Google has said since February 2023 that appropriate use of AI is not against its guidelines, and its October 2026 update added a fact-checking warning, not a penalty. What Google penalises is scaled content abuse — many low-value pages made mainly to manipulate rankings. On 16 October 2026, CLM Sequoia found 21.6% of top-ranking law firm pages were 76–100% AI-written, and their rankings held through the September 2026 spam update.

Can AI write content that follows Scots law?

Only if the system is configured for it. Generic AI defaults to English law — stamp duty instead of Land and Buildings Transaction Tax, English conveyancing instead of missives and Home Reports. Before buying, test the vendor with five Scots-law questions and read the answers. If any answer is English law, the system is not ready for an Edinburgh firm.

How much does AI content writing cost for a law firm?

ChatGPT Plus is $20 a month for unlimited drafts; Jasper Pro is $59 per seat a month on annual billing; legal-specific WebWriter.ai is $199 to $999 a month for 10 to 100 articles; humanised done-for-you content runs $300 to $800 an article. VisionQuantech's installed AI Content Writer is $2,500–$12,500 setup plus $500–$1,999 a month care, with 20% off setup for the first ten founding clients.

Do I need a solicitor to review AI-written content?

Yes. The ABA's Formal Opinion 512 (July 2024) makes lawyers responsible for AI-generated content published under their name, and the Law Society of Scotland's conduct rules carry the same duty in Scotland. More than 206 AI-hallucination cases reached court filings by mid-2025. Review is fifteen minutes per article — and it is the step that turns a draft into a publication.

What should a law firm never let AI write unsupervised?

Anything with a case citation, a statute reference, a fee or outcome promise, or a claim about the firm's experience. AI invents citations confidently, states English law as if it applied in Scotland, and writes guaranteed-outcome copy that breaks advertising rules. Those four categories always need a solicitor's eyes before publish.

References & further reading

  1. Clio — 2025 Legal Trends Report (79% of legal professionals use AI, up from 19% in 2023). https://www.clio.com/resources/legal-trends-report/
  2. Inoriseo — AI Legal Writer: Risks & Best Practices (Clio figures, 206 hallucination cases, 68% more leads). https://inoriseo.com/law-firm-seo/content-creation/ai-legal-writer/
  3. American Bar Association — Formal Opinion 512, Generative AI (29 July 2024). https://www.americanbar.org/groups/professional_responsibility/publications/ethics_opinions/aba-formal-opinion-512/
  4. Google Search Central — Guidance about AI-generated content (updated 1 October 2026: fact-check everything, including title tags and alt text). https://developers.google.com/search/docs/essentials/spam-policies
  5. Custom Legal / CLM Sequoia — One in Five Top-Ranking Law Firm Pages Is Almost Entirely AI (16 October 2026, 2,418 searches, 27.9% average AI content). https://custom.legal/knowledge/one-in-five-top-ranking-law-firm-pages-is-almost-entirely-ai-even-after-the-september-update/
  6. Lexicon Legal Content — Does Google Penalize AI Content? What Law Firms Need to Know. https://www.lexiconlegalcontent.com/does-google-penalize-ai-generated-content/
  7. AIMadeFor — AI Content Marketing for Law Firms (Ethical Guide). https://www.aimadefor.com/blog/ai-legal-marketing-content-lawyers/
  8. Deepsmith — Jasper Review: Features, Pricing, and Is It Worth It? (G2 4.7/5, Capterra 4.8/5, Trustpilot 3.2/5). https://deepsmith.ai/blog/jasper-review
  9. WebWriter.ai — AI Content for Law Firms (pricing $199–$999/mo). https://www.webwriter.ai/for/law-firms?from=juriscontent
  10. Inoriseo — ChatGPT for Law Firm Marketing (first-draft machine, not an SEO strategy). https://inoriseo.com/law-firm-seo/chatgpt-for-law-firm-marketing/

Cite this article: Shivam Kumar (2026). Can AI Write Content for Law Firms? An Edinburgh Firm's Honest Answer. VisionQuantech. https://visionquantech.com/blog/can-ai-write-content-for-law-firms-edinburgh.html

SHARE𝕏inf✆✈⧖

Related services & guides

Chat with us Email us
Advertisement