Let’s address the elephant in the room before we go on: you have been directed to this post by a law firm and if you have read some of our earlier Legal Loop email updates or attended one of Ryan’s Savvia.AI webinars and you have still been going to ChatGPT for legal advice, it begs the question: WHY?
On 29 October 2025, OpenAI published its updated Usage Policy and one change triggered a lot of commentary on social media channels with headlines such as: “ChatGPT can no longer give medical or legal advice”. A lot of people have an opinion on this and here is ours:
Firstly, has OpenAI just updated its policy or has it updated the technology to go into lockdown and give you the AI equivalent of the dreaded 404 error http code if a prompt seems remotely like it is the user seeking medical or legal advice?
Answer: it’s the policy that updated not the tech.
If a user inputs into ChatGPT a legal question, ChatGPT will still endeavour to provide an answer (as every good LLM should) BUT you will now see at the end an “important note” from ChatGPT telling you “this is general legal information, not legal advice” and then it goes on to say you if you are seeking to rely on the information you should verify it with a qualified lawyer. In other words, the onus is on the user to get a suitably qualified human to mark ChatGPT’s homework that you set it.
Now to the actual OpenAI Usage Policy update in question: in the previous version of the policy OpenAI said users should not use its AI for high-stakes decisions in sensitive areas without human review. Problem with that was not everyone knew or agreed on what that meant. So OpenAI has added clarity in their latest usage policy by explicitly prohibiting use of its services for:
- “provision of tailored advice that requires a license, such as legal or medical advice, without appropriate involvement by a licensed professional”; and
- “automation of high-stakes decisions in sensitive areas without human review” and it then lists sensitive areas as: critical infrastructure, education, housing, employment, financial activities & credit, insurance, legal, medical, essential government services, product safety components, national security, migration, law enforcement.
Our take on this is nothing has actually changed in reality in terms of best practice when using AI.
One of our key advice points on using AI responsibly in your business has always been to verify AI outputs and it seems obvious to say that if you want to ensure an AI output that is medical or legal in nature is accurate, the human you go to for that check is one who is trained (in the traditional human sense) in medical or legal advice whichever the case may be.
Going back to our question at the top as to why would you consider going to ChatGPT for medical or legal advice.
This was highlighted in the recent ‘Channel 4 Dispatches: Will AI take my job?’ The competitive edge that AI has over humans is speed and, consequently, cost. However, AI is (at the date of writing) superb for data quantity; it’s faster therefore it can do more in a shorter space of time, but it is not there with quality (yet) and that is where the human is still integral to responsible AI use.
By all means go to AI for information, but if you want to rely on what it tells you ask a qualified person to check it first. You might think AI is saving you time and money, however, it just takes a few AI hallucinations or bias to render its “advice” inaccurate which can undo your cost savings and more.
Just look at the Deloitte case in Australia in July 2025 where unchecked AI advice was given… if proper human checks had been applied and the hallucinations spotted earlier Deloitte could have saved itself $440,000.
Ready to adopt AI safely and confidently?
Our AI Compliance Package is your legal toolkit for safe, confident AI adoption. It gives you everything you need to harness AI’s potential without the legal headaches.
From tailored policies and contract clauses to hands-on guidance and expert tool reviews, we’ll help your business achieve compliance with current UK AI legislation while innovating responsibly – from just £1,500.
Start your AI journey the right way — book your Discovery Call today.
Kay graduated with a law degree from Southampton University and has over ten years experience practising commercial law in-house.
