How to recognise reliable security claims from cloud providers
Security claims are only reliable if you can verify them in documentation, audits and contracts; marketing language and logos are not evidence.
A reliable security claim is specific, verifiable and points to a source, such as technical documentation, an independent audit report or a data processing agreement. General words such as secure or military-grade say little without an explanation of key management and scope.
Who is it for: Individuals and small businesses who want to be able to tell the marketing language of cloud providers apart from evidence that you can check.
What is the difference between marketing language and a verifiable claim?
Marketing language is often meant to create trust: secure, private, heavily protected. A verifiable claim is one you can test. It says what exactly is protected, how, for which data and how someone has checked it. For example: files are encrypted on your device with a key that only the user has, and this is described in public documentation.
A useful tool is the following question for every claim: where can I read this, and what would I see if it were not true?
What counts as evidence?
Not all sources carry the same weight. Below is a rough order of indications. A data processing agreement is a contractual arrangement and does not prove how the technology works.
- Statements on the website or in adverts: a starting point, not evidence.
- Security documentation or a technical whitepaper: shows how the system is supposed to work.
- Open source code or a public description of the protocol: allows others to assess it.
- An independent audit report: confirms that a third party has tested the measures. Read the scope, the date and which parts are covered.
- A data processing agreement: records contractually what the provider does with your data and which sub-processors are used.
Which red flags are there?
Watch for signals that call for extra questions. One signal proves nothing, but several together are a reason to wait.
- Wording such as unbreakable, completely secure or military-grade, without technical explanation.
- A logo or certificate without mention of who issued it, what it covers and how current it is.
- Lack of clarity about who manages the key, or whether password recovery by the provider is possible while end-to-end encryption is promised.
- No information about metadata, sub-processors or where support works from.
- An audit report that covers only part of the service, while the claim is presented more broadly.
- A refusal to share documentation, while money is being asked for the service.
How do you check this in four steps?
You do not have to be a specialist to test the basics. Take the provider’s most important claim and go through these steps.
- Write the claim down word for word and highlight the vague words.
- Find the source: documentation, audit report or contract.
- Check scope and date: does the evidence cover this service, this plan and this feature?
- Ask the provider a question if something is missing and take the quality of the answer into account.
Translating claims into check questions
| Claim | Check question | Where to find it |
|---|---|---|
| Your data is encrypted | In transit, on the server or end-to-end, and who has the key? | Security documentation |
| We are certified | By whom, for which service and until when? | Audit report or the certificate itself |
| Your data stays in Europe | Including backups, support and sub-processors? | Data processing agreement and sub-processor list |
| We do not read your files | Are we technically unable to, or only not allowed to by policy? | Documentation on key management and access |
Who is this less suitable for?
- Anyone who just wants to make a quick choice: looking for evidence takes time and then yields little.
- Anyone who expects an audit report to guarantee everything: it gives a moment of testing, not complete certainty.
- Organisations with formal procurement requirements: have such an assessment carried out by a specialist or lawyer.
Frequently asked questions
Can I request an audit report as an individual?
Some providers publish a summary, others share it under conditions. Ask for it; a clear answer in itself says something.
Is open source automatically trustworthy?
No, but it allows others to check it. Also look at whether there are audits and whether the software is actively maintained.
What if a provider publishes nothing?
Then the claim remains unconfirmed. You can compensate by encrypting yourself, so that you depend less on the promise.
