Figures and guidance checked 22 July 2026.
Introduction
If you work in the mortgage industry or an associated sector such as surveying or conveyancing, “AI” can still sound like something confined to technology conferences and Silicon Valley boardrooms.
The reality is different. AI is already influencing clients’ expectations, competitors’ workflows and the systems used by lenders, regulators, professional bodies and public-sector organisations.
The question is no longer whether the industry should learn about AI. It is how quickly firms can build useful capability without creating a compliance, privacy or quality-control mess.
The global picture — investment is accelerating
Using their latest published guidance, Amazon, Microsoft, Alphabet and Meta expect combined capital expenditure of approximately $695bn to $725bn during 2026.
The companies use slightly different definitions and not every dollar is exclusively for AI. The spending also covers data centres, cloud capacity, chips, networking and other infrastructure. However, all four have explicitly linked much of the increase to demand for AI computing.
The wider investment figures are equally striking. Stanford University’s 2026 AI Index reported that global private investment in AI reached $344.7bn in 2025, an increase of 127.5% in one year. Generative AI companies attracted $170.9bn of that total, more than three times the amount invested during 2024.
Among organisations surveyed for the report, 88% said they were using AI in at least one part of their business during 2025. Generative AI was being used in at least one business function by 70%.
This infrastructure will not remain confined to technology companies. It is funding the tools that will increasingly be built into mortgage platforms, customer-management systems, document processing, underwriting, valuation and conveyancing workflows.
The UK picture — adoption is rising, but capability remains shallow
The latest Office for National Statistics research found that the proportion of UK businesses with 10 or more employees reporting the use of at least one AI technology increased from around 12% in late 2023 to around 35% in June 2026.
However, adoption is not yet particularly deep. Among businesses using AI, only 10% described their use as extensive. Just 11% of businesses said that more than half of their workforce had received AI-related training.
These figures are useful indicators of the wider business environment, although the ONS survey excludes finance and insurance and should not be treated as a direct measurement of the mortgage sector.
International worker-level research cited by the ONS also presents a more balanced picture than simply saying the UK is falling behind. Around 36% of UK workers reported using generative AI for their job in early 2026. That was below the US figure of 43%, but above the reported levels in France, Germany and Italy.
The UK is adopting AI. The more immediate problem is that adoption is uneven, training remains limited and many businesses are still experimenting without an organised strategy.
AI is already in our industry
This is not “future technology”. AI is already being used across property and financial services.
- Financial services: HM Treasury’s July 2026 Financial Services AI Adoption Plan reported that 21% of businesses in the broad finance and real-estate sectors had adopted AI in an early-2025 DSIT survey, compared with 16% across the economy. The latest published FCA and Bank of England survey, from 2024, found AI use among approximately 75% of participating regulated firms. An updated survey is under way.
- HM Land Registry: an AI tool developed by its own Data Science team was used to process local land-charge records for the London Borough of Newham. Work initially estimated to require 20 people for three months was completed in four weeks using four people. Every batch passed its data-quality assessment first time.
- Surveying: RICS’ first global professional standard for the responsible use of AI in surveying practice came into effect on 9 March 2026. It covers governance, risk management, professional judgement, procurement, output assurance and transparency with clients.
- Mortgage broking: an October 2025 Paradigm survey reported that 82% of respondents had used tools such as ChatGPT or Copilot during the previous three months, while 87% wanted to learn more. However, a separate June 2026 broker survey found that only around one-third were comfortable with greater AI or automation in areas such as document verification, affordability assessment, case triage and broker-lender communications.
The survey samples and methodologies differ, so these figures should be treated as indicators rather than a census of every mortgage firm. Nevertheless, the direction is clear: experimentation is widespread, while confidence in more consequential uses remains limited.
What happens if you don’t upskill
DSIT’s detailed AI Adoption Research, based on interviews with 3,500 businesses, found that the most commonly reported barriers were an inability to identify a worthwhile use case and limited AI skills or expertise.
When businesses assessed the seriousness of individual barriers, ethical concerns, cost and unclear regulation featured prominently. Data security and the accuracy of AI-generated outputs were the most common safety concerns.
For mortgage firms, the operational risks of failing to build capability include:
- Competitors completing administrative work and producing client communications more quickly.
- Higher processing costs where staff continue manually handling repetitive documents and data.
- Less consistent service and slower responses during busy periods.
- Employees using unapproved AI tools without suitable controls over client information.
- Management being unable to distinguish useful AI applications from expensive technology theatre.
Doing nothing does not necessarily prevent AI use. It may simply leave that use informal, invisible and unmanaged.
What AI can and can’t do in mortgages
AI should not be treated as a replacement for advisers, underwriters, surveyors or conveyancers. Its immediate value is in assisting with repetitive, document-heavy and administrative work so qualified people can spend more time applying judgement and dealing with clients.
What AI can do today:
- Transcribe meetings and produce draft summaries, fact-find notes and action lists.
- Extract and classify information from documents and flag missing or inconsistent items.
- Search approved lender criteria or internal knowledge bases, provided the system can identify and link back to its sources.
- Produce first drafts of client communications, internal checklists and marketing material for human review.
- Compare information across valuation, title, survey and conveyancing documents to prioritise further investigation.
- Convert unstructured information into consistent formats for case-management and compliance systems.
What it cannot safely do on its own:
- Guarantee that its answers, calculations or lender criteria are accurate and current.
- Fully understand the nuance of a client’s personal circumstances.
- Replace professional judgement on suitability, underwriting, valuation or legal matters.
- Make an unapproved platform suitable for processing confidential client information.
- Carry regulatory or professional accountability when something goes wrong.
AI can assist with the work. The adviser, surveyor, lender or conveyancer remains responsible for the outcome.
The compliance and governance angle
The regulatory position is becoming clearer, but it is not based on a separate rulebook labelled “AI”.
In June 2026, the FCA confirmed that it does not currently intend to introduce new AI-specific regulations. Instead, firms are expected to apply existing requirements, including the Consumer Duty, the Senior Managers and Certification Regime and established expectations around governance and controls.
The FCA is focusing on how firms:
- Oversee and govern their AI systems.
- Test models and monitor their outcomes.
- Protect customers, including people with characteristics of vulnerability.
- Explain decisions or outputs influenced by AI.
Data-protection duties also continue to apply. The ICO’s AI guidance addresses accountability, transparency, lawfulness, accuracy, fairness, security, data minimisation and individual rights. That guidance is currently under review following the Data (Use and Access) Act, so firms should check the current position rather than relying indefinitely on a copied policy.
For a smaller mortgage or property firm, proportionate governance might include:
- An approved list of AI tools and permitted business uses.
- A clear rule that identifiable client information must not be entered into unapproved or public AI services.
- Due diligence covering a supplier’s data retention, model-training, hosting, security, subcontracting and deletion arrangements.
- A named human reviewer for anything affecting clients, regulated activity or professional advice.
- Records of material AI systems, risk assessments, testing, incidents and any required data-protection impact assessments.
- Practical staff training covering both useful applications and prohibited behaviour.
Upskilling is therefore not only about productivity. It is part of maintaining effective systems and controls.
The opportunity for small firms
AI is not solely a big-bank opportunity. Smaller firms often have shorter decision-making paths, closer contact with their clients and fewer layers of legacy process.
A broker might use a controlled AI workflow to produce a clear, professionally reviewed mortgage summary shortly after a client meeting. A surveyor might automate the structure and formatting of a report while retaining full control over its observations, conclusions and recommendations.
No single improvement will transform a business. However, saving a few minutes at several stages of every case can reduce turnaround times, release staff capacity and improve consistency.
The objective should not be to remove people from the service. It should be to remove avoidable administration from the people providing it.
How to start — without spending big
You do not need a large transformation programme to begin.
- Choose one problem: start with a frequent, time-consuming and relatively low-risk task.
- Establish a baseline: record how long the task currently takes and where mistakes or delays occur.
- Test safely: use synthetic, anonymised or properly redacted information during initial trials.
- Check the tool: understand where data is stored, whether prompts are retained and whether customer information may be used to train the supplier’s models.
- Define human review: decide who verifies the output, what evidence must be retained and what the AI is not permitted to decide.
- Measure the result: compare time saved, error rates, turnaround times and customer outcomes rather than relying on impressive demonstrations.
Free or low-cost AI tools can be useful for learning with non-sensitive material. They should not automatically be used for fact-finds, identification documents, bank statements, medical information or other client data.
Free government-backed AI foundations training is also available through the UK’s AI Skills Boost programme, which aims to provide practical AI skills to 10 million workers by 2030, including at least two million SME employees.
A capability shift, not a software purchase
The biggest mistake is treating AI as a single product that can be purchased, switched on and forgotten.
Useful adoption requires people who can identify appropriate tasks, give the system clear instructions, recognise unreliable outputs, protect sensitive information and remain professionally accountable.
That capability develops through controlled experimentation and regular use. It does not develop from banning every tool, nor from giving staff unrestricted access and hoping for the best.
The firms most likely to benefit are those that start with real business problems, experiment safely, train their staff and improve their controls as the technology develops.
Conclusion
AI is no longer a buzzword floating around technology conferences. It is already influencing financial services, property data, surveying standards, mortgage systems and clients’ expectations.
In 2026, ignoring AI is both a strategic risk and a control risk. Rushing into it without governance is equally foolish.
The practical route forward is structured adoption: start with limited use cases, protect client information, check every important output and develop staff capability over time.
The question is not simply, “Will AI change the mortgage industry?” It is, “Which firms will learn to use it safely and effectively before the difference becomes obvious to their clients?”
