Cybersecurity & Data Protection  ·  22 September 2026

The Bill the Client Couldn't Read

A German court found TLS insufficient for a €15,000 invoice. Encrypt it properly, the way the vendors selling the fix want you to, and the client can't open it at all.
By Alan Wright  ·  The Haunted Lighthouse Limited  ·  Peel, Isle of Man

A construction firm in Schleswig-Holstein sent a final invoice to a private client by email: a PDF, TLS-encrypted in transit, nothing unusual about it. Criminals intercepted it en route, altered the bank details, and forwarded the manipulated version on. The client paid the fraudsters. The builder demanded payment again. The client refused, and the case went to the Oberlandesgericht.

In December 2024, the court sided, in part, with the client. TLS transport encryption, it found, was not an adequate security measure under Article 32 GDPR for an invoice carrying this much financial risk. End-to-end encryption was required. The builder kept his original claim against the client, but the client won a damages claim of equal size under Article 82 GDPR for the data protection breach, which offset the debt to zero.

Since then, a small industry of encrypted-email vendors has built a marketing campaign on top of this ruling. The pitch is straightforward: the courts have spoken, TLS is dead, here is our product. What that pitch leaves out is worth a closer look, because the actual compliance path the ruling implies runs into a wall almost immediately.

The reductio

Take the ruling at face value and do the responsible thing. Encrypt the invoice properly, end-to-end, the way the vendors are selling it. Send it to the client.

The client cannot open it. They have no PGP key, no S/MIME certificate, nothing to decrypt an asymmetrically encrypted attachment with, because true end-to-end encryption is not something you can just switch on for a recipient who has never set up key material and has no reason to know what any of this means. A private homeowner paying a heating contractor is not going to generate a keypair on request.

So the compliant version of this transaction produces an invoice the client cannot read. The court wanted risk-proportionate security. What a literal reading of the judgment produces, applied without thought, is a document nobody can act on.

What the ruling actually said

It is worth being precise about the case itself, because the marketing gloss compresses a fact-specific, risk-weighted finding into a blanket rule.

The court's own reasoning (OLG Schleswig, judgment of 18 December 2024, case 12 U 9/24) acknowledges that the GDPR itself gives no clear standard for what security measures are required. The appropriate level, the court says, has to be determined by weighing the relevant interests, with the legitimate security expectations of ordinary business dealings, tempered by reasonableness, playing a non-trivial role. That is a proportionality test tied to the facts of one case, not a new statute. The €15,000 sum at stake was doing real work in that weighing; a modest invoice on the same reasoning would not obviously trigger the same outcome.

Two things sit alongside that finding and rarely make it into the LinkedIn posts.

First, this standard is not new. The Datenschutzkonferenz, the joint body of Germany's federal and state data protection authorities, published guidance on exactly this question in May 2021, over three years before this judgment. That guidance draws precisely the line the court later applied: transport encryption is an adequate baseline for normal-risk email; where a breach of confidentiality would pose a high risk to the data subject, the guidance says end-to-end encryption is regularly required, alongside qualified transport encryption. The 2024 ruling did not invent a standard. It attached liability, for what appears to be the first time, to a standard that had been sitting in official guidance, unenforced in practice, since 2021.

Second, the case the vendors never cite didn't reach GDPR at all, and that is the more interesting problem for their pitch, not a footnote. OLG Karlsruhe, 27 July 2023, case 19 U 83/22: a business sold a used car to another business for €13,500, invoiced by email, and a hacker intercepted the exchange and substituted fraudulent bank details, the same mechanism exactly. But this was two companies dealing with corporate payment information, so GDPR's Article 32 duty never engaged at the jurisdictional threshold in the first place; Article 2(1) and Article 4(1) restrict the regulation's scope to personal data belonging to natural persons. With no GDPR duty in play, the court fell back to ordinary German civil law, a contractual duty of care under section 241(2) BGB, and measured it against established commercial expectations (Verkehrskreise) rather than any encryption-specific standard. It found no such duty to encrypt. Leave to appeal (Revision) was not granted. The seller was not liable. The buyer paid twice.

That is not two courts splitting on the same question. It is two different statutory regimes answering two different questions, and the answer changed because the applicable law changed, not because judges disagreed about what "good enough" security looks like. Which makes the omission worse for the vendor pitch, not better: the entire "courts now require E2EE" argument only holds inside a narrow lane, personal data, a natural person on the other end, a GDPR duty actually engaged, and evaporates the moment a transaction sits outside it. A vendor selling encrypted invoicing as a general business necessity is quietly assuming every prospective customer's invoicing carries GDPR-regulated personal data at the sender's own compliance duty. True for consumer-facing invoices. Not obviously true for a great deal of ordinary B2B correspondence.

The DPA already thought about the client who can't decrypt anything

Here is the part that actually complicates the reductio above, and it is worth being precise about it, because getting this wrong would overstate the vendors' cynicism in the wrong direction.

The 2021 DSK guidance is, on its face, entirely about email: transport encryption standards, when qualified TLS suffices, when end-to-end encryption over SMTP is required. It does not fold a web portal into that framework as a form of encrypted email. It treats the portal as a different channel altogether; footnote 4 raises it in the context of reaching a natural person the sender cannot otherwise establish secure email with, and the fix on offer is not "encrypt the email better." It is "stop using email for this and hand the person a browser session instead."

That distinction matters for what the vendor pitch is actually doing. The portal-and-password products sold into this anxiety, Tuta among them, are not implementing the DSK's email-encryption standard through a clever workaround. They are implementing the DSK's fallback for when email-based delivery cannot meet that standard, and then marketing the result as "encrypted email," which is not really what it is. The underlying compliance gap the DSK flagged in 2021, unencrypted high-risk correspondence to consumers, is real, and the portal is a legitimate, DPA-endorsed answer to it. But it is an answer that leaves email behind rather than one that secures it, and four years of vendor copy calling this "email encryption" has blurred that line to the point where it is worth restating plainly: if your invoicing solution routes the client to a login page rather than decrypting an attachment in their inbox, you have not encrypted your email. You have stopped emailing them the invoice.

Integrity and confidentiality are not the same problem

One more mismatch worth naming, because it is the kind of thing that gets flattened when lawyers write about cryptography. The actual harm in the Schleswig-Holstein case was an integrity failure: the invoice was altered in transit, and the altered version is what did the damage. The court's remedy, drawn from Article 32 GDPR, is confidentiality-flavoured: it is about who can read the data, not whether the data was tampered with.

A cryptographic signature, PGP or S/MIME, verifies that a document has not been altered since the sender signed it. It does nothing to hide the contents from anyone who intercepts it. Encryption hides the contents. It does not, on its own, prove the document wasn't swapped for a different one before encryption. Signing alone would have caught this specific fraud outright: the manipulated PDF would have failed signature verification the moment the client's mail client checked it. The court's stated remedy, encryption, does not directly address the mechanism of harm at all. It is the right general-purpose hygiene, but it is solving an adjacent problem, not the one that actually produced the €15,000 loss.

Where this lands for the Isle of Man

This is a German ruling, interpreting EU GDPR, from a regional appellate court. It does not bind Manx courts, and the Isle of Man's own data protection regime runs under the DPA 2018, aligned to UK GDPR, not EU GDPR. Nothing here is directly enforceable against a Manx firm as a matter of precedent.

But plenty of Island firms, particularly the fiduciary, legal and finance sectors this publication watches most closely, deal with German counterparties, German clients, or German-domiciled beneficiaries as a matter of routine. If your invoice, your engagement letter, or your data-carrying correspondence is landing in a German inbox, and the underlying transaction carries meaningful financial risk to a private individual on the other end, the exposure is practical rather than jurisdictional. Nobody is going to cite Schleswig-Holstein case law against a Manx company in a Manx court. Somebody might well cite it in correspondence, in a dispute, or in a German counterparty's own risk assessment of doing business with you, and by then the underlying question, was the transport-layer security actually adequate for the risk, is the same question regardless of which court would eventually hear it.

The sane response is not to buy an encrypted-invoicing subscription because a vendor's LinkedIn post said to. It is to apply the same risk-tiered thinking the DSK laid out in 2021: ordinary correspondence, transport encryption is fine; anything carrying high financial exposure to a private individual, particularly cross-border, warrants a harder look at the channel, and a second, out-of-band verification step for any change to payment details, which would have stopped both the Schleswig-Holstein and the Karlsruhe fraud regardless of what encryption was or wasn't in place.

A note on sourcing

The full text of the Schleswig-Holstein judgment, hosted on the state's juris-based judicial portal, was not accessible while researching this piece. The portal's own OAuth handshake is caught in a redirect loop on its own infrastructure, confirmed by direct inspection of the request chain rather than assumed; the fault sits with the portal, not with any client-side configuration. Citations to the judgment's specific paragraphs (Rn. 67 and surrounding text) are drawn from dejure.org's paragraph-indexed excerpts rather than the primary document. The Datenschutzkonferenz guidance was retrieved and read in full from the regulator's own published PDF. The Karlsruhe judgment is cited via dejure.org and corroborating secondary legal commentary. A second, independent fact-check pass confirmed the Rn. 67 characterisation against separate secondary commentary and flagged the original framing of Karlsruhe as a direct doctrinal split with Schleswig-Holstein as imprecise; that correction, the two judgments sit under different statutory regimes rather than disagreeing on the same question, is reflected in the text above. Readers with direct juris or beck-online access are better placed than this piece to quote the Schleswig-Holstein text verbatim; what's presented here is accurate to the paragraph-level citations available, not to a full read of the primary document.


Sources


Editor's note: this piece applies the analytical framework set out in "The Theatre Pulldown", examining the gap between institutional governance claims and operational reality.

Questions about this analysis, or interested in working with The Haunted Lighthouse?
consultancy@haunted.lighthouse.co.im

The Sovereign Auditor covers digital sovereignty, cybersecurity governance, and data protection policy, with particular focus on Isle of Man jurisdiction and Crown Dependency issues.

Support independent analysis. Subscribe directly, or scan on your phone.

Payments via PayPal. Credentials delivered by email. No Substack. No Stripe. No middlemen.