Did the incorrect Legal 500 testimonials appear in AI answers?
We found no evidence that they did. We reviewed 705 AI answers to barrister searches recorded between 1 and 3 October 2026, and none reproduced the incorrect testimonial material. The window was short and the site was offline for part of it, so this shows no evidence of spread, not proof that it could not happen.
Key takeaways
- None of the 705 AI answers to barrister searches reproduced the incorrect Legal 500 testimonials. The window was short, so this is no evidence of spread, not proof.
- AI leans on rankings most when asked who is best or who to instruct. Legal 500 was used in 43% of those answers (467 of 1,090).
- The accuracy of a ranking source matters to every firm AI describes, so firms should check what AI says over weeks, not once.
What did we find?
This article draws on the same 1,645 AI answers as our Legal 500 research, collected between 4 September and 3 October 2026. The barrister check is separate.
- Barrister check: 705 AI answers to barrister searches, recorded between 1 and 3 October 2026.
- Result: none reproduced the incorrect testimonial material.
- Limit: the window was short, and the Legal 500 site was offline for part of it.
- Context: Legal 500 was used in 43% of best, top and instruct answers (467 of 1,090), and in 7% of specialist answers (40 of 555).
What is Legal 500, and why does AI use it?
Legal 500 is an independent legal directory that publishes rankings of law firms and barristers. AI uses it as a source when a search asks it to compare.
In our data, Legal 500 was used in 31% of all answers (507 of 1,645). For best, top and instruct searches it was 43% (467 of 1,090). We describe the wider pattern in our Legal 500 research.
What happened at Legal 500?
On Thursday 1 October 2026, Legal 500 published the latest edition of its annual United Kingdom (Bar) rankings. The next day, it took its website offline after critical testimonial material appeared on barrister profiles.
Legal 500 said an error in the upload process during a website update meant a small amount of testimonial information was published incorrectly. It said it removed the content as soon as it became aware, and that it would strengthen its quality assurance.
We have not reproduced any of the material here. This article does not comment on how Legal 500 runs its process, or on the merits of any ranking.
Why does an error at a ranking source matter to AI answers?
An error at a ranking source matters to AI answers because AI leans on independent rankings most when a search asks who is best or who to instruct. In our data, Legal 500 was used in 43% of best, top and instruct answers (467 of 1,090), against 7% of answers asking who specialised in an area (40 of 555). For best and instruct searches, 56% of the 14,433 pages AI read were directories, rankings and lists.
That makes ranking sources part of how AI describes a firm. When AI leans on a source, the accuracy of that source matters to every firm it describes.
Did the error reach AI answers?
In the 705 AI answers we checked, the incorrect testimonials did not appear. We reviewed 705 AI answers to barrister searches recorded between 1 and 3 October 2026, and none reproduced the incorrect testimonials.
The window was short and the site was offline for part of it. This is no evidence of spread. It is not proof that spread could not happen.
| Item | Detail |
|---|---|
| Answers checked | 705 AI answers to barrister searches |
| Dates | 1 to 3 October 2026 |
| Result | None reproduced the incorrect testimonial material |
| Limits | Short window. The Legal 500 site was offline for part of it |
| What it shows | No evidence of spread. It is not proof that spread could not happen |
Table 1. The barrister check at a glance.
Why does a short window prove so little?
A short window proves little because AI answers move from day to day. In our main dataset, the first firm AI named for the same search changed from one day to the next 56% of the time (664 of 1,193 consecutive-day comparisons across ChatGPT, Perplexity and Google AI Overviews). We look at this in more detail in why one AI check tells a law firm almost nothing.
Our barrister check covered three days. It cannot show what AI said before the check began, or what it will say next week. That is why we describe the result as no evidence of spread, and not as proof.
Who checks the sources AI relies on?
In our view, checking what AI says about a firm falls mostly to the firm itself. Directories have always shaped reputation. What has changed is the route. A ranking or profile can now be read and summarised by an AI system for a prospective client who never visits the directory.
In our view, that makes the accuracy of a firm’s third-party profile a visibility issue as well as a reputation one.
The timing adds to the point. Ahrefs data shows UK searches for ‘legal 500’ peak every October, at more than 14,000 in both October 2024 and October 2025, around two and a half times the August level. In our view, the weeks after publication are when an error in a ranking source is most likely to be seen.
What should a firm or chambers do?
A firm or chambers should treat what AI says about it as something to monitor, not a one-off check:
- Check what AI says about the firm or chambers across several platforms, over weeks rather than once.
- Look at which sources each answer drew on.
- Check the firm’s profiles on directories, LinkedIn and regulator registers for accuracy and consistency.
- Keep a dated record of any wrong answer, with the platform and the source it used.
- Raise errors with the source that published them.
My view: the check is the easy part
The first question I would ask after an incident like this is whether the damage has travelled. In my view, few firms can answer that, because few are measuring what AI says about them at all.
Our result is reassuring but thin. Three days and 705 answers is a snapshot, which is why I will not call it proof.
What concerns me more is the route. A client can now ask AI who to instruct, and a directory can shape the answer without the client ever visiting the directory. A firm cannot control that source. It can check it, and it can watch what AI says over time. I would put that check in the diary next to the annual directory submission.
We run this with every client we work with, as it’s so important and as multiple sources now get cited, you need to stay on top of it for compliance and reputation management.
Sean O’Dwyer Curran
Methodology
- Period: 4 September to 3 October 2026 for the main dataset. The barrister check is separate: 1 to 3 October 2026, 705 answers.
- Barrister check: AI answers to barrister searches recorded with Searchable. A testimonial counted as reproduced
- Main dataset: 1,645 AI answers. 15 high-intent legal searches. 10 asked who was best, top, leading, recommended or who to instruct. 5 asked who specialised in, or had expertise in, an area.
- Practice area: mainly high-value family law and family Bar searches, London and UK-wide.
- Platforms: ChatGPT, Google AI Overviews and Perplexity, daily. Gemini and Copilot, daily from 25 September 2026.
- Data: AI answers collected with Searchable. Search demand and site traffic estimates from Ahrefs (UK).
- Definition of ‘used’: the source appeared among the pages the AI system accessed to generate its answer, whether or not it was shown as a visible citation. An answer counts once, however many pages from the same site it used.
- Day-to-day change: the first firm named for the same search, compared across consecutive days on ChatGPT, Perplexity and Google AI Overviews (1,193 comparisons).
- Limitations: the barrister check covers a short window, during part of which the Legal 500 site was offline. The main searches focus on family law. AI answers vary from day to day.
Frequently asked questions
Did the incorrect Legal 500 testimonials appear in AI answers?
We found no evidence that they did. We reviewed 705 AI answers to barrister searches recorded between 1 and 3 October 2026, and none reproduced the incorrect testimonial material. The window was short and the site was offline for part of it.
Why did Legal 500 take its website offline?
Legal 500 said an error in a website update caused a small amount of testimonial information to be published incorrectly. It took the site offline the day after publishing its latest Bar rankings. It said it removed the content when it became aware, and that it would strengthen its quality assurance.
Can an error at one ranking source spread through AI answers?
Our check does not answer that. We found no evidence of spread in 705 answers over a short window. A longer check, across more searches, would be needed to say more.
Does AI copy Legal 500’s rankings?
Not directly. AI used Legal 500 pages in 31% of the answers we analysed, usually alongside other sources such as Chambers and Partners and law firm websites. Using a page as a source is not the same as repeating its rankings.
What can a firm do if AI says something wrong about it?
Record the answer with the date, the platform and the sources it drew on. Then ask the source that published the error to correct it. Fabled Labs’ AI Recommendation Audit shows which sources shape AI answers for a firm and its competitors.
Final word
AI has not replaced the legal directories. It reads them, which makes their accuracy part of every firm’s visibility. In the 705 answers we checked, the incorrect testimonials did not appear. The firms best placed are those that check what AI says about them over time, not once.
Sources
- Legal Cheek, ‘Legal 500 apologises after publishing incredibly harsh reviews of barristers in website update error’, October 2026.
https://www.legalcheek.com/2026/10/legal-500-apologises-after-publishing-incredibly-harsh-review-of-barristers-in-website-update-error/ - Non-Billable, ‘Legal 500 takes website offline after critical barrister reviews appear’, 2 October 2026.
https://www.nonbillable.co.uk/news/legal-500-takes-website-offline-critical-barrister-reviews - City AM, ‘Legal 500 takes down site after accidentally publishing damning barrister reviews’, October 2026.
https://www.cityam.com/legal-500-takes-down-site-after-accidentally-publishing-damning-barrister-reviews/ - RollOnFriday, ‘Legal500 accidentally publishes terrible reviews of top barristers’, October 2026.
https://www.rollonfriday.com/news-content/exclusive-legal500-accidentally-publishes-terrible-reviews-top-barristers - Fabled Labs analysis of AI answer data collected with Searchable, 4 September to 3 October 2026. Barrister check: a separate set of 705 answers, 1 to 3 October 2026.
- Ahrefs Keywords Explorer, UK search volume history for ‘legal 500’, accessed 3 October 2026.
This article reports research by Fabled Labs based on AI answers collected between 4 September and 3 October 2026. AI outputs change frequently, and results may differ for other searches, locations and dates. Legal 500 and Chambers and Partners are independent publications. Fabled Labs is not affiliated with either, and this article does not comment on the merits of any ranking or firm. This article is for information only and is not legal advice.