Why the UK Financial Ombudsman Is Ditching AI Templates for Real Phone Calls

Why the UK Financial Ombudsman Is Ditching AI Templates for Real Phone Calls

Have you ever read a consumer complaint that sounded like a Supreme Court brief, only to find out the cited laws don't actually exist?

That exact nightmare has been flooding the UK's Financial Ombudsman Service (FOS). Consumers and claims management companies (CMCs) have been using generative AI tools like ChatGPT to draft formal complaints against banks and insurers. The result is an avalanche of hyper-formal, bloated letters filled with confident legal jargon and outright AI hallucinations.

Instead of building better chatbots to fight back, the FOS made a counterintuitive move. Investigators are picking up the phone. By calling complainants directly early in the process, the ombudsman is cutting through thousands of words of synthetic fluff to find out what actually went wrong.

It's a brilliant, human-first solution to a very modern mess.

The Anatomy of an AI-Generated Bank Complaint

If you've ever worked in dispute resolution or compliance, you know what a genuine consumer complaint looks like. Usually, it's messy. People write in plain English, sometimes with typos, expressing frustration because a bank froze their account, mis-sold them car finance, or denied a travel insurance claim. You can hear their actual voice.

Now compare that to what the FOS started receiving en masse over the last couple of years.

Suddenly, everyday consumers were submitting twenty-page documents packed with references to obscure Consumer Duty clauses, fabricated Financial Conduct Authority (FCA) handbook rules, and nonexistent case law. Generative AI doesn't just polish grammar. When prompted to "write a strong legal complaint against my bank," large language models invent regulatory breaches to satisfy the user's request.

Here is why that breaks the system:

  • Investigators have to check everything. An ombudsman can't just ignore a cited statute. Even if an investigator suspects a consumer used AI, they still have to verify whether the quoted regulation applies or if it's pure fiction.
  • The core issue gets buried. AI loves padding. A simple dispute over a £150 overdraft fee gets wrapped in five pages of philosophical waffle about "unfair contractual imbalance" and "systemic fiduciary failure."
  • Expectations become wildly unrealistic. When a chatbot tells a consumer their bank violated three federal laws (sometimes even citing US laws in a UK complaint), that consumer expects massive compensation. When the FOS explains the bank actually followed UK rules, trust evaporates.

Honestly, you can't blame regular people for trying. Filing a formal financial complaint feels intimidating, and staring at a blank page is hard. Typing a two-sentence prompt into a free AI tool feels like leveling the playing field against a giant bank's legal department.

Except it does the exact opposite. It actively slows down their own payout.

Why Claims Management Companies Made It Worse

Individual consumers using ChatGPT is only half the story. The bigger headache for the Financial Ombudsman Service comes from professional claims management companies and high-volume representatives.

Under the FOS's current fee structure, businesses face case fees when complaints are referred to the service, and the ombudsman has recently introduced charges for professional representatives who bring high volumes of poorly evidenced claims. Before those fee changes kicked in—and even after—some CMCs saw generative AI as a way to industrialize complaint filing.

Instead of interviewing a client properly to gather bank statements, dates, and specific losses, certain reps started feeding basic intake forms into AI scripts. Those scripts spat out massive, templated complaints designed to overwhelm bank complaints teams and push cases toward the FOS.

I've seen how this plays out behind the scenes in financial operations. When a bank receives a generic, AI-padded letter that doesn't actually attach proof of loss, the bank rejects it. The rep then automatically escalates it to the ombudsman. The FOS case file arrives bloated with text, yet completely empty of the actual evidence needed to make a fair decision.

You end up with a gridlocked queue where genuine, urgent cases—like someone locked out of their mortgage funds—sit stuck behind thousands of copy-pasted AI hallucinations.

How Picking Up the Phone Solves the Hallucination Problem

When an investigator calls a complainant directly, the AI smokescreen collapses in about ninety seconds.

Think about what happens on a live phone call. An investigator doesn't ask, "Please elaborate on paragraph fourteen regarding your assertion of statutory misrepresentation." They ask, "Tell me in your own words what happened when you tried to make that transfer."

People don't hallucinate on the phone. They tell you the truth. They say, "I got a text that looked like it was from my bank, I clicked the link, and £2,000 vanished from my savings."

Suddenly, the investigator has everything they actually need. They know it's an Authorized Push Payment (APP) fraud case. They know the exact amount. They know how the scam started. Those eighteen pages of AI-generated legal theory become completely irrelevant, saving hours of wasted administrative reading and back-and-forth emails.

What Early Phone Contact Actually Accomplishes

The FOS's push toward early telephone engagement isn't just about catching fake citations. It fundamentally changes the psychology of dispute resolution.

First, it strips away the adversarial armor. Written complaints—especially AI ones—are combative by default. A calm, human conversation reminds the consumer that the ombudsman is an impartial referee, not an enemy who needs to be beaten with legal jargon.

Second, it identifies missing evidence immediately. On a ten-minute call, an investigator can tell a consumer exactly which three screenshots or statements will win their case, rather than sending formal information requests that take three weeks to bounce back.

Third, it manages expectations early. If a consumer has been misled by a chatbot—or a shady claims firm—into thinking they're owed £10,000 for minor administrative inconvenience, an investigator can gently explain how FOS awards actually work. Many people accept a fair, realistic resolution on the spot once someone actually listens to them.

The Bigger Lesson for Customer Service and Legal Ops

Every major industry right now is obsessed with throwing AI at incoming volume. Banks use AI to triage complaints; consumers use AI to write complaints; lawyers use AI to summarize both. It's a closed loop of machines talking to machines while actual humans get further away from a resolution.

The UK Financial Ombudsman Service is proving that when synthetic text floods the zone, human voice becomes the ultimate verification tool.

You can fake a legal brief in four seconds. You can't fake a coherent, five-minute conversation about your own lived experience without the cracks showing immediately.

If you're a business owner, a compliance lead, or someone running a customer support operation, pay attention to what the FOS is doing here. If your team is drowning in overly formal, weirdly structured emails or support tickets that don't quite make sense, stop trying to parse the text. Pick up the phone. A three-minute call will save you three weeks of email ping-pong.

How to File a Financial Complaint That Actually Wins

If you're a consumer preparing a complaint against a bank, lender, or insurer—whether in the UK with the FOS or anywhere else—ditch the AI legal generator. Investigators hate it, and it hurts your credibility.

Follow this exact approach instead if you want your money back fast:

  1. Write like a normal human being. Use plain, everyday English. Investigators are trained to apply the law to your facts; you don't need to quote statutes or regulations to prove you were wronged.
  2. Stick to a strict timeline. List what happened in chronological order with exact dates. "On March 12, I called customer support. On March 15, the payment left my account." Dates win cases; adjectives don't.
  3. State the exact financial loss. Don't ask for vague "damages." Specify the exact pound or dollar amount you lost, any interest or overdraft fees directly caused by the bank's mistake, and any concrete impact on your daily life.
  4. Attach raw proof, not summaries. Provide screenshots of text messages, PDFs of bank statements, and call logs. Let the documents speak for themselves.
  5. Answer the phone when the investigator calls. If your case reaches the ombudsman, welcome that early phone call. Be honest, stick to the facts you experienced, and let them guide you on what specific evidence will close your file.

Keep your complaint simple, factual, and unmistakably human, and you'll move straight past the AI noise to the front of the queue.

MT

Mei Thomas

A dedicated content strategist and editor, Mei Thomas brings clarity and depth to complex topics. Committed to informing readers with accuracy and insight.