The Quantum Horizon: Why Harvest Now, Decrypt Later Changes Everything

The biggest risk of the quantum revolution is not just the future collapse of encryption. It is that the damage is already happening. Because classical encryption like RSA and Diffie-Hellman forms the backbone of the modern internet, adversaries are currently intercepting and stockpiling data for future decryption. This Harvest Now, Decrypt Later approach means the security of past communications is already compromised, regardless of when a functional quantum computer arrives. Readers who distinguish between securing the future and protecting the past gain an advantage in managing long-term information risk. This shows that the five-year horizon for quantum computing is not a deadline for action, but a window that has already begun to close.

The Hidden Cost of Fast Cryptographic Solutions

Turing Award winners Charles Bennett and Gilles Brassard clarify a dangerous misconception about post-quantum cryptography. While many organizations rush to implement software-based defenses, Bennett and Brassard explain that these are not the same as quantum-hardened solutions. Post-quantum cryptography relies on mathematical assumptions that certain problems will remain difficult for quantum computers.

> Post quantum crypto by definition can be broken again with unlimited computing power. Why it is quantum crypto cannot? And moreover, post quantum crypto is based on beliefs. It is really a wishful thinking that what they are inventing will indeed be unbreakable with a quantum computer.
>
> -- Gilles Brassard

Relying on these solutions creates a false sense of security. Because these systems are classical, they remain vulnerable to future algorithmic breakthroughs. In contrast, quantum cryptography uses physics-based key distribution to provide a guarantee based on the laws of nature rather than the limits of computing power. The advantage belongs to those who recognize that solved does not mean secure.

The Harvest Now, Decrypt Later Trap

The most overlooked dynamic is the permanence of intercepted data. Most engineers focus on the moment of decryption, but the real threat is the storage of encrypted traffic.

> The reason we should panic, the main reason, is that it is known as the harvest now and decrypt later paradigm by which nothing prevents an opponent of intercepting all cryptid communication and decrypt everything, which rectively.
>
> -- Gilles Brassard

This creates a feedback loop where current digital infrastructure acts as a permanent archive for adversaries. Once information is harvested, it is effectively public property waiting for the clock to run out on current encryption. This changes the incentive for long-term data protection: if your data has a shelf life of diplomatic or personal sensitivity, the risk is not the future. It is the present.

Redefining the Human-AI Relationship

Bennett reframes anxiety about AI by using the theory of interactive proofs. He suggests that our role is not to compete with the machine, but to act as Arthur, the limited but honest participant, interacting with Merlin, the infinitely wise but untrustworthy machine.

You do not need to trust the AI to benefit from it. By learning to ask adaptive, challenging questions, you can force the machine to prove its work. This turns the black box of AI into a tool for verification rather than a source of truth. The advantage lies in developing the skill to audit the machine rather than blindly consuming its output.

Key Action Items

• Audit Data Sensitivity (Immediate): Identify which communications or data assets have long-term sensitivity of 10 years or more. Assume that anything currently encrypted with RSA or Diffie-Hellman is already compromised in the eyes of state-level actors.
• Decouple Future-Proofing from Current Infrastructure (Next Quarter): Stop assuming that switching to post-quantum software solutions solves the harvest now problem. These are stopgaps, not permanent fixes.
• Adopt an Arthurian AI Workflow (Ongoing): Shift your interaction with AI from generating answers to verifying proofs. Treat AI as an untrustworthy expert and design your prompts to require the AI to explain its reasoning or provide evidence you can audit.
• Practice Information Hygiene (Immediate): As Bennett notes, the only truly effective protection for information is not saying things you would regret if they became public. This remains the most durable security protocol.
• Monitor Quantum Hardware Progress (12-18 Months): Watch for the hockey stick of technological progress in quantum computing. The shift from theoretical to physical realization is accelerating, so do not rely on outdated five-year projections.