Security Claims for Money Apps: Say What You Can Prove
Security · 10 min read ·
Which security statements a money app can make honestly, how to word them, and which popular slogans to drop because nobody can check them.
"Bank-grade security." "Military-grade encryption." "Your data is 100 per cent safe." Open a dozen finance app pages and you will see these phrases again and again. They sound reassuring. They are also close to meaningless, because no standard defines them, nobody can test them and each reader imagines something different.
A careful reader, and a money product attracts careful readers, quickly learns to distrust such slogans. A better approach is to say what you actually do, in specific terms you can back up. This guide shows how to turn security from a slogan into a short list of facts.
Why vague claims backfire
There are three problems with slogans.
They cannot be checked. If nobody can say what bank-grade means, a reader cannot judge it, and you cannot prove it.
They set a trap. "100 per cent safe" is an invitation to be proven wrong, and when something goes wrong, the gap between promise and reality becomes the story.
They may break advertising rules. The UK advertising regulator's guidance on substantiation says marketers must hold documentary evidence for claims that consumers are likely to regard as objective. A claim like "unhackable" is likely to be read as objective, and it cannot be substantiated.
Specific statements work better on all three counts. They are testable, honest about limits and credible to experts.
A vocabulary of honest claims
Here are categories of claims a money app can make, with examples of wording. Use only those that are true for you.
Sign-in and accounts
- "You can turn on two-step sign-in. We recommend you do."
- "We lock an account after repeated failed sign-in attempts and tell you by email."
- "Passwords are stored as one-way hashes, not in readable form."
Say what is optional and what is mandatory. If two-step sign-in is optional, do not imply it is on by default.
Data in transit and at rest
- "Connections to our site are encrypted."
- "Sensitive data is encrypted when stored."
Keep it short. You do not need to name algorithms on a launch page, though a security page for technical readers can. What matters is that the claim is true for all the data you describe, not only some.
Money and card details
- "Card details are handled by a payment provider certified to the PCI DSS standard; they never touch our own servers."
The PCI Security Standards Council publishes the card industry's data security standard and describes the range of standards it maintains. If you take or process card data, your obligations depend on how you do it; most small products reduce them by using a certified provider. Say that you use one, and which kind of arrangement it is, only if it is accurate.
Access to your data
- "Our staff cannot see your bank password because we never receive it."
- "A small number of named staff can access customer data, and every access is logged."
If you read data through a regulated connection or an official interface, say so and link to the explanation.
Testing and review
- "An independent firm tested our systems in [month and year]. We fixed the issues it found."
- "We run a vulnerability disclosure process; report issues at the address below."
If you say you have been tested, state who, when and roughly what the scope was. Keep the report, even if you cannot publish it.
Certifications and audits
State the name of the standard, what it covers, the date and who issued it. Do not display a badge you do not hold. UK advertising rules state that trust marks and approvals must not be used without authorisation, and the regulator has ruled against firms that did so.
If you are working towards a certification, say "working towards", not "compliant".
Incident response
- "If we find a problem affecting your data, we will tell you by email as soon as we can and say what to do."
A calm incident plan is a security feature. Write one even if you are small.
What to leave out
- Slogans without content: "bank-grade", "military-grade", "ultra-secure".
- Absolutes: "unhackable", "completely safe", "zero risk".
- Borrowed credibility: logos of banks or regulators you do not work with.
- Jargon salad: long lists of acronyms with no explanation.
- Detail that helps attackers: exact versions, internal architecture or the names of tools in use.
Use a framework to organise your thinking
You do not need to invent your own list of controls. The US National Institute of Standards and Technology publishes a Cybersecurity Framework that organises security work into functions such as identify, protect, detect, respond and recover. The UK's National Cyber Security Centre publishes practical guidance for organisations of different sizes. Using a recognised structure to review your own practice makes it easier to spot gaps, and easier to explain to others what you do.
A simple exercise for a small team: write one sentence under each function describing what you do today. Where you cannot, you have found your next task.
Write a security page
A single page that states your practices honestly is one of the best trust signals you can offer. Include:
- A short summary in plain English.
- The controls above that are true for you.
- How to report a vulnerability, with a monitored address.
- What users can do to protect themselves.
- The date the page was last reviewed.
The NCSC's top tips for staying secure online include things like using strong passwords, turning on two-step verification and keeping software updated. Linking to advice like that shows you care about users' own safety as well as your own.
Help users protect themselves
Many losses come from phishing and weak passwords, not clever attacks on your servers. Help users:
- Explain how your real emails look and what you will never ask for, such as a full password.
- Offer two-step sign-in and show how to set it up.
- Send alerts for sensitive changes, such as a new device or a changed email address.
- Provide a quick way to lock an account.
These are all things you can describe on the launch page in a sentence each.
Be honest about limits
No honest firm can promise that nothing will ever go wrong. Say what you do to reduce risk, what you do if something happens and what you cannot control, such as a user's own device. Honest limits make the rest of the page more credible.
For example: "No system is perfectly secure. We reduce risk with the controls above and keep improving them. If something goes wrong, we will tell you quickly."
Handling the first incident
Sooner or later, something will happen. A phishing email using your name. A bug that exposes a screen. A provider outage. Prepare:
- Decide who decides. One person leads, one communicates.
- Know your obligations. Data protection rules in many countries require reporting some breaches to a regulator and, in some cases, to the people affected, within set times. Check what applies to you.
- Tell people early. An early, honest, incomplete message is better than a late, perfect one.
- Say what to do. Change passwords, watch for odd messages, contact you.
- Publish what you learned. A short, factual write-up builds more trust than silence.
Security and AI features
If your app uses AI to analyse transactions or answer questions, say what data it sees, where it is processed and whether it is used to train models. The OWASP project publishes a list of top risks for applications built on large language models, which is a useful checklist if you add such features. Be specific about what the model can and cannot do, and never let it act on money without clear user confirmation.
A claims check before you publish
For each security sentence on your page, ask:
- Is it true for everything it implies?
- Can I show evidence if asked?
- Would an expert find it meaningful?
- Does it give an attacker useful detail?
- Is it dated, if it can go stale?
Delete or change anything that fails.
On this site
The privacy and terms pages explain how data and listings are handled here, and the contact page is the route for reporting a concern. The launches page shows how products present themselves, and the submit page starts an entry.
A sample security summary
To show how plain this can be, here is a short summary a small team might publish. "We never see your bank password; account data arrives through an authorised connection. Your own password is stored only as a one-way hash. You can turn on two-step sign-in in settings, and we recommend it. Card payments are handled by a certified provider, so card numbers never reach our systems. Two people on our team can access customer records, and each access is logged. An independent firm tested our systems earlier this year, and we fixed everything it rated serious. To report a problem, write to the address below and a human will answer. We reviewed this page on the date shown."
Every sentence is checkable, none is boastful and the whole thing takes under a minute to read. It also gives your team a standing checklist: if any line stops being true, the page must change the same day.
Keep the evidence folder
Behind the page, keep a folder with the proof: the testing summary, certificates, supplier assurances, the incident plan and the date each was last reviewed. When a journalist, partner or careful customer asks, you can answer in minutes. When your own team changes, the folder keeps knowledge from walking out of the door.
The short version
Drop the slogans. Say specifically how sign-in, data, card details and staff access work, who tested you and when, which certifications you truly hold and what you do after an incident. Publish a dated security page, give users tools to protect themselves and admit that no system is perfect. Facts a sceptic can check beat any adjective.
Questions and answers
- Is 'bank-grade security' an acceptable claim?
- It is vague and cannot be measured, so many reviewers treat it as a warning sign. Describe the actual controls instead.
- Should I publish the results of security testing?
- A short summary stating who tested, when and what scope is useful. Never publish details that would help an attacker.
- What certifications matter?
- Only those you actually hold and that match what you do. State the name, scope and date, and be ready to show the evidence.
- How do I explain two-step sign-in to users?
- In two sentences: what it is, how to turn it on. Put the link where people sign in.
- What should I say after a security incident?
- Tell affected people promptly, say what happened and what to do, and publish an honest account once you know the facts.