Facts at a Glance
AI & Mortgage Advice: What the Research Shows
AI models wrong 57% of the time on financial questions · Error rate rises to 88% on complex questions · Tested across 18 leading AI models, 10,000+ responses (Saturn, 2026) · FCA: general-purpose AI tools are not FCA-regulated
AI can answer a mortgage question in seconds. It can explain mortgage terms, estimate borrowing figures and even talk you through different types of mortgages in a way that sounds clear and confident.
There is an important question to ask before relying on that answer: is it actually right for you?
According to a new study from Financial Reporter, Saturn Research found that AI models gave incorrect answers to financial questions 57% of the time, with the error rate rising to 88% on more complex questions. The research tested 18 AI models across 121 financial questions, with each question repeated several times and more than 10,000 responses assessed.
Saturn serves financial advice firms and has also called for AI-generated financial advice to come under FCA regulation, so its commercial interests and position on AI regulation are worth keeping in mind when considering the findings.
The research does not mean that every answer from AI is wrong, or that AI has no place in financial research. It does show why you should be careful when an AI-generated answer involves an important financial decision.
A mortgage is particularly personal. Your income, employment, deposit, credit history, existing commitments, property and future plans can all affect what may be available to you. Different lenders can also assess the same circumstances in different ways.
AI can give you information. The important question is whether that information actually applies to your circumstances.
This article looks at what the latest research tells us about AI and financial advice, where AI can be useful when researching a mortgage, and why personalised mortgage advice still matters.
What Did the New AI Financial Advice Research Actually Find?
The Saturn figures give us one reason to be cautious about relying on AI for financial questions. There is also research from the Financial Conduct Authority looking at how large language models perform when they are used to provide consumer financial information.
In one FCA research project, researchers tested GPT-3.5 and GPT-4 in two different areas. The first looked at whether AI could simplify complicated financial concepts and explain them in a way that was easier for consumers to understand. The second compared AI-generated responses with a traditional website-based question and answer format for cash savings queries.
The FCA found that large language models showed strong potential for simplifying complex financial information and making it easier to understand. That is one of the areas where AI can be genuinely useful. Someone who does not understand a mortgage term, for example, could use AI to get a simpler explanation before doing further research.
The FCA also found that the usefulness of AI was context-dependent. The way an AI tool is built into the customer journey, the information it is given and how its responses are evaluated can all affect the outcome. The FCA concluded that checking AI outputs requires a combination of automated testing and human judgement.
This is an important distinction for anyone researching a mortgage. AI can be useful for helping you understand a subject or giving you a starting point for your own research. That does not mean the answer should automatically be treated as personalised financial advice.
There is also a wider issue around how much people trust AI when making financial decisions. In August 2026, separate FCA research found that 56% of UK investors aged 18 to 40 who currently invest or are considering investing said they trusted AI tools, while 44% mistakenly believed AI-generated financial information was regulated. The FCA also found that 38% believed it was acceptable to make an investment decision based solely on AI output.
The research was about investment decisions rather than mortgages, so the figures should not be presented as evidence about mortgage borrowers specifically. It does, however, show why the distinction between AI-generated information and regulated financial advice is becoming increasingly important.
Taken together, the research points to a more balanced view. AI can be useful for learning and exploring financial topics, although its answers still need to be checked and put into context. When the question becomes specific to your own mortgage and financial circumstances, that distinction becomes particularly important.
Why Can AI Get Mortgage Questions Wrong?
Mortgage questions can look simple on the surface. A question such as “How much can I borrow?” might seem straightforward. In reality, a lender needs to consider a range of information before deciding what it may offer.
Mortgage Criteria Are Complicated
A mortgage application can involve income, employment, deposit, credit history, existing debts, regular expenditure, property type, loan-to-value (LTV), mortgage term and other personal circumstances. Each lender also has its own criteria.
Two people earning the same salary can receive different borrowing figures depending on their wider circumstances. It is not always as simple as applying a standard income multiple.
AI can explain some of these factors and help you understand what they mean. The problem comes when an answer does not take every relevant detail into account, or when the information it uses is incomplete or out of date.
Lenders Do Not All Use the Same Criteria
Another important point is that there is no single set of mortgage criteria used by every lender.
One lender may be more comfortable with a particular type of employment or income, while another may take a different approach. This can be especially relevant for people with circumstances that do not fit a straightforward employed application.
That means an answer based on a general mortgage rule may not reflect what is actually available from the lenders relevant to your circumstances.
Mortgage Rules and Criteria Can Change
Mortgage criteria can change over time. Lenders can update their approach to affordability, income, employment, property types and other parts of their lending criteria. Wider rules and guidance can also change.
AI may not always have access to the latest information, particularly when you are asking about a specific lender or a recent change in mortgage criteria.
For this reason, AI can be useful as a starting point when you want to understand a mortgage topic. Important information should still be checked against current and reliable sources.
When the question changes from “What does this mortgage term mean?” to “What mortgage could I get based on my circumstances?”, the answer needs to go beyond general information and take your situation and the relevant lender criteria into account.
The Biggest Problem With AI Mortgage Advice: Your Mortgage Is Personal
One of the biggest limitations of AI mortgage advice is that a mortgage is not simply a question with one correct answer. What may be appropriate for one homebuyer may not be appropriate for another, even when their basic financial figures look similar.
An AI tool can only work with the information it is given. It may produce a confident answer based on a simplified version of your situation, while a proper mortgage assessment can involve details that are harder to capture in a short question.
The Context Behind the Numbers Matters
Your financial circumstances do not exist in isolation. Your employment situation, existing financial commitments, plans for the property and what you want the mortgage to achieve can all change the options worth considering.
This is particularly important when your circumstances are not straightforward. A general answer may make sense in one situation but be less relevant once the wider details are considered.
Suitability Is More Than Finding a Possible Mortgage
There is also a difference between finding a mortgage you could potentially qualify for and understanding whether it is suitable for your circumstances.
You may be able to borrow a certain amount, for example, but that does not automatically mean you need to borrow the maximum available to you. The mortgage term, monthly payments, interest rate and your plans for the future can all be relevant when considering your options.
This is where personalised mortgage advice differs from a general AI response. The question is not simply “What mortgage can I get?” It is also “What should I consider when deciding which mortgage is right for my circumstances?”
AI can help you understand the questions. Personalised advice is about looking at the answers in context.
Can AI Tell You How Much Mortgage You Can Borrow?
AI can give you an estimate of how much you might be able to borrow. This can be useful when you are at the early stages of looking at properties and want to get a rough idea of the price range you could consider.
The important distinction is between an AI estimate and a lender’s affordability assessment.
For example, you might tell an AI tool that you earn £45,000 a year, have a £30,000 deposit and have no outstanding debts. It may use those figures to produce an estimated borrowing range.
That estimate is not the same as knowing what a lender would actually agree to.
Lenders use their own affordability calculations and lending criteria when assessing an application. The result can therefore vary between lenders, even when the basic figures entered into a calculator remain the same.
If you are at the early stages of planning your purchase, our How Much Can I Borrow Calculator can give you an initial indication of your potential borrowing.
A calculator is still only a starting point, just like an AI estimate. Your actual borrowing will depend on your circumstances and the criteria of the lender that considers your application.
The useful question is therefore not only “How much does AI say I can borrow?” but “What could I realistically borrow based on my circumstances and the lenders available to me?”
What Can AI Actually Help With When Researching a Mortgage?
AI can be useful when you are learning about mortgages, especially if you are starting from scratch and do not know what questions to ask.
For example, you can use AI to:
- Explain mortgage terms and terminology in simpler language
- Help you understand the difference between fixed-rate and tracker mortgages
- Give you a basic explanation of how deposits and loan-to-value ratios work
- Suggest questions you may want to ask a mortgage adviser
- Help you organise the information you need before making an enquiry
- Explain general mortgage concepts before you look for more detailed information
This can make the early stages of mortgage research easier. If you come across a term you do not understand, asking an AI tool for a plain-English explanation can be a useful first step.
The important distinction is between using AI to understand information and using AI to make a personalised mortgage decision.
AI can help you learn about your options and prepare questions. When you need to know which mortgage options are actually suitable for your circumstances, you need information that takes your individual situation and current lender criteria into account.
What Can AI Not Reliably Replace in Mortgage Advice?
AI can help you understand mortgages, compare general information and prepare questions. There are parts of the mortgage process where general information is not enough.
Assessing Your Full Circumstances
A mortgage adviser can look at your circumstances as a whole rather than relying on a few figures entered into a tool. This can include your income, employment, commitments, deposit, credit history, property plans and any other details that may affect your application.
The more complicated your circumstances are, the more important those details can become.
Checking Current Lender Criteria
Lenders do not all assess applications in exactly the same way. An adviser can research current lender criteria and identify which lenders may be relevant to your circumstances.
This can be particularly useful if your situation does not fit a standard mortgage application, such as having variable income, being newly self-employed, owning several properties or looking at a property with unusual requirements.
Considering Suitability
Finding a mortgage that you could potentially qualify for is not necessarily the same as finding an option that suits your circumstances.
An adviser can discuss the available options with you, explain the differences and help you understand the wider implications of the mortgage you are considering.
Supporting the Application
Mortgage advice does not necessarily end once you have found a potential mortgage. There can be documents to provide, questions from the lender to answer and different stages to work through before the mortgage completes. Having someone who understands your application can be useful throughout that process.
AI can help you prepare and understand what is happening. It does not have the same role as a mortgage adviser who is reviewing your circumstances, researching lender options and supporting you through an actual mortgage application.
AI vs a Mortgage Broker: What Is the Difference?
AI and a mortgage broker can serve very different purposes when you are researching a mortgage.
AI can therefore be useful for learning and preparing. A mortgage broker has a different role when you need advice based on your individual circumstances and want to move towards an actual mortgage application.
What Does the FCA Say About AI and Financial Advice?
The Financial Conduct Authority (FCA) is also looking closely at how consumers are using AI for financial information and decision-making.
In its 2026 Mills Review, the FCA looked at how AI could change retail financial services over the coming years. The review recognises that AI can make financial information easier to access and understand, while also identifying risks around misleading information, consumer harm, fraud, bias and people relying too heavily on AI when making financial decisions.
General-Purpose AI Tools Are Not FCA-Regulated
There is an important distinction between using a general-purpose AI tool and receiving regulated financial advice.
The FCA says that general-purpose AI tools such as ChatGPT and Claude are not set up specifically to provide financial advice. When consumers use these tools for financial decision-making, they are not receiving regulated advice and do not currently have the same Financial Ombudsman Service or Financial Services Compensation Scheme protections that can apply within the regulated financial services framework.
That does not mean you should avoid AI altogether. It means you should understand what you are getting when you ask a general-purpose AI tool a financial question.
Consumers Are Already Putting Trust in AI
The FCA’s recent research into younger investors also shows why this matters.
In a July 2026 survey of 666 UK adults aged 18 to 40 who either already owned investments or were considering investing, 56% said they trusted AI tools for investment information. The FCA also found that 44% mistakenly believed AI-generated financial information was regulated, while 38% believed it was acceptable to make an investment decision based solely on AI output.
This research looked at investment decisions rather than mortgages, so it should not be treated as evidence about mortgage borrowers specifically. It does, however, show the wider issue the FCA is examining: consumers may place more trust in AI-generated financial information than they should.
For someone researching a mortgage, the practical lesson is fairly simple. AI can be a useful way to learn and prepare questions, but you should not assume that a confident-sounding answer is regulated financial advice or that it takes responsibility for the decision you make.
Should You Stop Using AI for Mortgage Research?
There is no need to stop using AI altogether when you are researching a mortgage. The useful approach is to understand what AI can help with and where you need more reliable, personalised information.
AI can be helpful when you are:
- Learning basic mortgage terminology
- Exploring questions you have about buying a home
- Understanding the difference between different mortgage types
- Preparing questions for a mortgage adviser
- Organising information before making an enquiry
For more important questions, it is worth checking the information against reliable sources and current lender criteria rather than taking an AI response at face value.
Why Mortgage Advice Still Needs a Human Element
AI can explain a lot about mortgages, but there is still an important role for human advice when you are making a decision about your own finances.
A mortgage adviser can look at the details behind the numbers, ask questions about your circumstances and consider information that may not fit neatly into an AI prompt. They can also research current lender criteria and explain why certain mortgage options may or may not be suitable for you.
At Everest Mortgages, we are an independent, whole-of-market mortgage broker. This means we are not tied to one lender and can search across the market for mortgage options that fit your circumstances. Our advisers are qualified and FCA regulated, with experience helping clients with a wide range of mortgage situations.
There is also a practical side to mortgage advice. Once you have decided to apply, there can be paperwork, lender questions and different stages to work through. Having an adviser who understands your application can make that process easier to manage.
AI can give you information and help you ask better questions. Personalised mortgage advice is about understanding your circumstances, researching the available options and helping you through the process of making a mortgage decision.
How Everest Mortgages Can Help
If you have been using AI to research your mortgage, the next step is to move from general information to advice based on your own circumstances.
At Everest Mortgages, we start by understanding your situation, including your income, employment, deposit, existing commitments and what you are hoping to achieve. We can then research mortgage options across the market, explain the differences and help you understand what may be available to you.
If you are buying your first home, our First Time Buyer Guide can help you understand the main stages of buying a property and getting a mortgage. If you are still working out your budget, our How Much Can I Borrow Calculator can give you an initial indication of your potential borrowing.
We also help clients with different mortgage needs, including remortgages, self-employed mortgages, bad credit mortgages and buy-to-let mortgages. We also work with clients looking at specialist options such as teacher mortgages, where lender criteria can vary.
You can also find out more about John Everest and his experience as a mortgage adviser.
Whether you are just starting your research or are ready to apply, we can help you understand your options and support you through the mortgage process.
Frequently Asked Questions About AI Mortgage Advice
Still have questions about using AI when researching a mortgage? Here are some of the most common questions homebuyers ask, with straightforward answers to help you decide when AI is useful and when you may need more personalised information.
Can ChatGPT give mortgage advice?
ChatGPT can provide general information about mortgages and help you understand terminology, calculations and common questions. For a decision about your own mortgage, however, you need information that takes your circumstances and the current mortgage market into account.
Can AI tell me how much mortgage I can borrow?
AI can give you an estimated figure based on the information you enter. It cannot guarantee what a lender will agree to, so an estimate should be treated as an indication rather than a confirmed borrowing amount.
Is AI mortgage advice reliable?
AI can be useful for general research, although its answers can sometimes be inaccurate, incomplete or out of date. It is sensible to check important information before using it to make a financial decision.
Can AI compare mortgage lenders?
AI can explain differences between lenders and mortgage products based on the information available to it. It may not have the latest product details or complete lender criteria, so its comparison should not be treated as a definitive view of the market.
Is AI financial advice regulated in the UK?
General-purpose AI tools are different from regulated financial advisers. The FCA has raised concerns about consumers misunderstanding the status of AI-generated financial information, so it is important to know whether the information you are receiving is regulated advice or general information.
Should I use AI or a mortgage broker?
You can use AI to help with your initial research, learn terminology and prepare questions. A mortgage broker becomes more relevant when you want to understand your own mortgage options and move from general research towards an application.
Can AI replace a mortgage adviser?
AI can take care of some useful research tasks, but mortgage advice involves more than answering questions. Understanding your circumstances, considering available options and helping you through the mortgage process all require a more personalised approach.
What information should I check before applying for a mortgage?
Make sure the information you are using is current, particularly figures relating to affordability, interest rates, lender criteria and mortgage products. You should also have a clear picture of your income, deposit and regular financial commitments before applying.
Thinking About Your Next Mortgage?
AI can be a useful part of your mortgage research. It can help you understand the terminology, explore questions and get a better idea of what you need to consider. When you are ready to make a decision, personalised advice can help you understand what may be available based on your circumstances.
Thinking about buying or remortgaging? Speak to Everest Mortgages and book your free consultation to discuss your circumstances and your next steps.
