The digital signature that is binding today may not prove anything tomorrow. The challenge of digital signatures in the face of post-quantum cryptography
- Pablo Peñaranda
- 2 days ago
- 5 min read
Imagine a contract signed digitally this week. Everything is in order: a valid certificate, a timestamp, and the other party identified beyond a doubt. In 2026, that document is legally binding. The uncomfortable question is another: Will it still be binding in 2035? What if, by then, anyone with access to a quantum computer can forge a signature identical to yours on a text you never approved?
Almost all the discussion about the quantum threat revolves around encryption to protect secrets.
In this very blog, we’ve mapped out the risk and the clock ticking behind it (“An X-Ray of Quantum Risk in Colombia”: https://www.cyte.co/post/radiografia-riesgo-cuantico; “A Radar for Q-Day”: https://www.cyte.co/post/q-day-seguridad-de-la-informacion).
It’s a necessary conversation, but it leaves out the area that will hit Colombian companies the hardest first: digital signatures. And it will hit them hard not because of what it reveals, but because of what it fails to prove.
A digital signature is, at its core, a legal promise
In Colombia, the digital signature is not merely a technological nicety. Law 527 of 1999 granted data messages the same legal validity as paper documents, and Decree 2364 of 2012 regulated electronic signatures on that basis. Much of daily operations hinges on this: electronic invoices that the DIAN requires to be signed and kept intact from the moment they are issued, as well as contracts, minutes, and employment records.
That validity rests on three attributes: authenticity (who signed), integrity (that nothing was altered), and non-repudiation (that the signer cannot later claim, “It wasn’t me”). What is almost never said out loud is that all three are, ultimately, a mathematical assertion. Non-repudiation does not rest on an abstract legal principle, but on a technical fact: that no one other than the holder could have produced that signature. The day that fact ceases to be true, the attribute loses its foundation and with it, the confidence that a signed document is binding on the person who signed it.
The first pillar to fall is non-repudiation
Of the three attributes, the quantum threat affects one more than the others. Not confidentiality. Not integrity. Non repudiation. It’s important to distinguish between two issues that people tend to confuse.
Encryption suffers from the classic “harvest now, decrypt later” problem: capturing encrypted traffic today to decrypt it tomorrow—a strategy I’ve already detailed in “Cryptography as an Attack Surface” (https://www.cyte.co/post/criptografia-algoritmo-como-superficie-de-ataque). It’s bad, yes, but it’s a confidentiality issue: the real damage occurs the day the key is compromised.
The signature has a different problem and, for many businesses, a worse one. RSA and elliptic curve (ECDSA, which underlies almost every certificate in Colombia) signatures rely on mathematical problems that Shor’s algorithm solves on a sufficiently large quantum computer. Solving those problems not only allows one to read the signature it also allows one to sign it. With the private key reconstructed, an attacker can generate signatures indistinguishable from legitimate ones. And since non-repudiation relies on the fact that this was previously impossible, there is no longer any technical way to prove that a signed document was signed by its owner. The algorithm, which was supposed to provide security, becomes a vulnerability.
It's not just a risk going forward. It's retroactive. Everything you've signed and filed remains, on Q-Day, just as questionable as anything signed afterward.
A signature doesn't protect for just a moment—it protects for years
A TLS channel encrypts a session that lasts only a few seconds; when it ends, the value of that key plummets. A signed document, on the other hand, must remain verifiable throughout its entire lifespan. And that lifespan is long: business records are kept for years, a long-term contract may need to be verified long after it has expired, and an employment obligation may be disputed decades later. Validity isn’t needed on the day it’s signed. It’s needed on the day someone challenges it which is usually the worst possible moment.
That time lag between when I sign and when I have to prove it is exactly the window that the quantum threat exploits. And it isn’t closed by migrating the encryption.
Migrating the framework isn't the same as changing a library
Three practical steps:
First, know where you’re signing. Most organizations don’t have a clear inventory of where they use ECDSA or RSA: how many certificates, in which integrations, and with what expiration dates. You can’t migrate what you don’t know exists.
Second, adopt hybrid schemes and timestamping. Signing with both a classical algorithm and a post-quantum algorithm ensures that the signature remains valid even if one of the two fails. And for items already archived, re-stamping them with a post-quantum timestamp provides proof that they existed and were valid before the old algorithm was compromised.
ML-DSA and SLH-DSA: there is no single post-quantum signature.
The good news is that the standard already exists. In 2024, NIST published the first post-quantum signature schemes, and it did not select just one for a reason worth understanding.
ML-DSA (FIPS 204), based on the Dilithium scheme, operates on lattices. It’s fast, and its signatures are a manageable size, making it the natural candidate for day-to-day volume: billing, authentication, and online transactions.
SLH-DSA (FIPS 205), formerly known as SPHINCS+, is built solely on hash functions. Its signatures are large and it’s slower, but its security rests on a very conservative and well-studied assumption. For documents that need to remain credible thirty years from now—such as a notarized record or an infrastructure contract—that underlying robustness outweighs a few extra kilobytes.
The lesson isn’t “pick one.” It’s that long-term and high-volume firms have different needs, and for the first time, there are tools tailored to each.
Third, treat certificates and keys as an asset to be managed, not as something you issue and then forget about. The ability to rotate algorithms, reissue certificates, and respond without slowing down crypto-agility—in short, that’s what separates an orderly migration from a race against the clock. (https://www.cyte.co/post/el-camino-hacia-la-resiliencia-poscuantica-y-la-criptoagilidad).
At Cyte, we take this approach: Notyfind automatically discovers the certificates distributed throughout the organization, monitors their expiration dates, and automates their renewal and reissuance—so that when it’s time to change the algorithm, you don’t have to go searching, one by one, for where each signature was located.
Two migrations, two different issues at stake
Migrating encryption protects what you want to keep secret. Migrating the signature protects something even harder to recover: your word, and the legal validity of everything you’ve already signed. The first problem becomes apparent the day your data is stolen. The second becomes apparent the day someone—with all the technical justification in the world says, “That signature proves nothing.”
The encryption you abandon today will cost you a secret. The signature you abandon today will cost you the proof of your word. Better not to let that day come.
Comments