Ethereum Researcher Says AI May Break Encryption Before Quantum Does

We have spent the last decade worrying about the wrong monster under the bed. For years, the prevailing nightmare in crypto has been the 'Quantum Apocalypse'—the idea that one day, a massive quantum computer will blink into existence and instantly turn our private keys into public property. We have built entire roadmaps around 'Quantum Resistance' as if it were the final boss of cryptography.

But Justin Drake, a researcher at the Ethereum Foundation, recently shifted the conversation in a way that should make every founder and developer pause. He is not worried about the hardware race. He is worried about the software. Specifically, he thinks Artificial Intelligence might figure out how to break our encryption through raw mathematical intuition before a stable quantum computer even finishes cooling down its processors.

The Math Problem Nobody Wants to Face

To understand why this matters, you have to look at what encryption actually is. It is not magic; it is just a very hard math problem. Current systems like ECDSA, which secures your Ethereum and Bitcoin, rely on the assumption that it is computationally impossible to work backward from a public key to find a private key. It is like trying to un-drop a glass of water and expecting the droplets to land back in the cup in the exact same configuration.

The traditional view is that to break this, you need brute force. You need a machine that can try billions of combinations at speeds we cannot currently achieve. That is the quantum argument. But AI changes the equation. AI does not just 'try' things; it looks for patterns. It finds shortcuts. If there is a hidden symmetry in the way we generate these keys—a flaw in the math that humans have missed for forty years—AI is the perfect tool to find it.

Drake’s warning is simple: AI could potentially 'solve' the math of elliptic curve cryptography. If an LLM or a specialized neural network discovers a non-quantum way to reverse these functions, the hardware requirements drop to almost zero. You would not need a billion-dollar laboratory in a basement at IBM; you would just need a high-end GPU cluster.

The Immediate Risk: Address Reuse

This is where it gets practical for builders and users. Drake is not just talking about theory; he is giving tactical advice that sounds suspiciously like common sense we have ignored for too long. His main suggestion? Stop reusing addresses. Move your funds to fresh, unused addresses.

Here is why that matters from a technical perspective. When you have a fresh Ethereum address that has never sent a transaction, your public key is not actually fully visible on the blockchain. It is hashed. Hashing is a one-way street that is significantly harder to break than the signature itself. It is only when you sign a transaction and broadcast it to the mempool that you reveal your raw public key to the world.

If you are a founder holding a treasury or a developer building a wallet, this is a wake-up call. We have become lazy with 'persistent identity.' We like the idea of one ENS name, one address, and one history. But that persistence is a vulnerability. If AI cracks the math behind the public key, any address that has ever sent a transaction becomes a target. The ones that have just sat there, receiving funds but never sending, remain shrouded behind the hash.

What This Means for the Roadmap

For those of us building in this space, Drake’s perspective forces a rethink of the 'long-term' roadmap. If the threat is AI math rather than Quantum hardware, the timeline for 'going dark' or moving to ZK-STARKs (which are generally considered more resistant to these types of attacks) just moved up by a decade.

We cannot wait for the hardware to arrive. We have to assume the software is already being trained. Think about the resources being poured into OpenAI, Anthropic, and DeepMind. These companies are building machines designed to recognize patterns in data that humans cannot perceive. Cryptography is, at its core, a set of patterns. We are betting the entire global digital economy on the hope that these patterns are truly random. AI is the ultimate tool for proving they are not.

Takeaway for Builders

• Prioritize Privacy by Default: If your protocol encourages users to keep funds in the same address for years while signing frequent transactions, you are building a legacy liability.
• Adopt Modern Primitives: We need to stop treating Post-Quantum Cryptography (PQC) as a '2030 problem.' If AI can optimize the math, 2030 is actually next Tuesday.
• Rotation is Security: We need better UX for address rotation. The reason people reuse addresses is that managing twenty keys is a nightmare. This is an opportunity for smart contract wallet builders to automate the safety measures Drake is suggesting.

A Skeptic’s Silver Lining

Now, I am a skeptic by nature. Is it possible this is just another way to generate urgency in a market that thrives on fear? Sure. But there is a fundamental difference between 'The sky is falling' and 'The math might be weaker than we thought.' Drake is not telling you to sell your ETH; he is telling you to move it to a fresh wallet. That is a low-cost, high-reward piece of hygiene.

The real takeaway here is that we have spent too much time focusing on the physical limitations of computers and not enough on the evolution of intelligence. A smarter brain (AI) is often more dangerous than a faster hammer (Quantum). As builders, our job is to assume the hammer is already swinging and the brain is already thinking. If you are holding significant assets in an address that has been active for years, it is time to cycle. Not because a quantum computer is coming tomorrow, but because a very smart algorithm might already be looking at your public key.

Honesty in this space is rare, and Drake is giving us a dose of it. The math we rely on is a choice, not a law of physics. If the choice is compromised, the only defense is to stay one step ahead of the disclosure. Move your funds, use new addresses, and stop assuming that yesterday's encryption is tomorrow's fortress.

Read the original at Decrypt →