Data stolen today, decrypted later: which of your data it affects

Someone can capture encrypted data today, store it, and wait. If a capable quantum computer arrives, data protected with today’s public-key methods could be decrypted then. The G7’s roadmap for the financial sector notes that data “may be at risk even if it is intercepted well before” such a machine exists. Security teams call this “harvest now, decrypt later”.

It’s the reason quantum risk is a planning question today, even though nobody can say when such a machine will exist.

How the attack works

When two systems talk securely, they usually use public-key cryptography to agree a shared secret key, and then a symmetric method to encrypt the conversation with it. The symmetric part is strong. The weak point, against a future quantum computer, is the public-key step that agreed the key.

An attacker who records the whole exchange today can keep it. If that public-key step can later be broken, the attacker can recover the session key and read everything that was protected by it. Nothing has to be broken today for the data to be exposed tomorrow; it only has to be captured.

How long your data must stay private

What matters is how long your data has to stay private, more than when quantum computers arrive. A password reset link loses its value in minutes. A medical record, a long-term contract or a person’s financial history may need to stay private for decades.

A simple test helps. Add the number of years a piece of data must stay private to the number of years it will take your organization to move its protection to quantum-safe methods. If that total reaches into the period when a capable quantum computer might exist, the data you send today is already exposed, because it will still matter when it can be read.

Nobody can give that period a date, and the test doesn’t need one. It shows which data is most exposed relative to the rest, which is what you need to set priorities.

Which data matters most

Start with data that must stay confidential for a long time and travels over networks:

• Health and research records.
• Long-term contracts and intellectual property.
• Personal financial records and identity data.
• Legal, government and state records.

Data that loses its value quickly matters much less here. So does data that never leaves a protected environment, although “never” is worth checking: backups, replication and partner connections move more data than most diagrams show.

A worked example

Illustrative: an insurer sends claims data, including medical details, between its own data centres and a partner that processes payments. The records must stay private for decades. The connection uses strong encryption today, with keys agreed using an elliptic-curve method.

Applying the test, the claims connection ranks near the top: the data lives a long time, and moving the connection means coordinating with a partner, which will take time. A daily price feed on a similar connection ranks near the bottom, because its data is public within hours.

Questions to ask about your own data

• Which of our data must stay confidential for ten years or more?
• Where does that data travel over networks, including backups and partner connections?
• How are keys agreed on those connections today?
• Which of those connections depend on someone outside our organization to change?

What to do

Protect what must stay private longest first. That means knowing where that data travels, what protects it on the way, and which systems would need to change to protect it with quantum-safe methods.

Start with how keys are agreed in transit, because that’s the step a recorded conversation depends on. NIST’s first quantum-safe standards were published in 2024, so the replacements exist today, and NIST describes hybrid approaches, running a quantum-safe method alongside a traditional one, as a way to move without breaking compatibility.

The broader sequencing, including systems whose signatures need to stay trusted, is covered in a deadline isn’t a migration plan.