Bitcoin's long-term plan for surviving quantum computers has always been built on a candid assumption: some of today's cryptographic algorithms will break. The only question is which ones, and when. A recent demonstration involving the HAWK digital-signature scheme and an AI model from Anthropic has made that question far more immediate.

The HAWK system, a proposed post-quantum signature protocol, is not deployed anywhere, and Bitcoin does not use it. Yet it sits in the same quantum race that Bitcoin developers are scrambling to finish. The reason is simple: the cryptographic foundation underneath Bitcoin, and most of the modern internet, is not designed for the arrival of powerful quantum machines.

Key facts in this report

• Anthropic's Claude Mythos Preview model discovered a flaw in the proposed HAWK digital-signature scheme, effectively halving its smallest key strength.
• The AI-driven attack took about 60 hours and $100,000 in computing, reducing the work needed to break HAWK's smallest parameter set from roughly 2^64 to 2^38 operations.
• HAWK is a candidate for replacing current web and banking signatures in a post-quantum world, but it is not yet deployed anywhere.
• Bitcoin and ether signatures are unaffected today, but the result reinforces warnings that classical cryptanalytic attacks are quickly improving.
• Bitcoin developers are debating how and when to migrate to quantum-resistant cryptography, and this event adds urgency to that debate.

What the AI-assisted attack found

Anthropic's Claude Mythos Preview model uncovered a flaw in HAWK that effectively halves the strength of the scheme's smallest key size. The attack, which ran for about 60 hours and consumed roughly $100,000 in cloud computing, reduced the number of operations required to break HAWK's smallest parameter set from 2^64 to 2^38. In cryptographic terms, that is a staggeringly large reduction. A 2^64 obstacle is considered borderline possible for well-resourced attackers; 2^38 is closer to a speed bump for a determined organization with specialized hardware.

The discovering model did not rely on quantum computing at all. It used classical cryptanalysis techniques—looking for mathematical shortcuts, structural weaknesses, or exploitable patterns in the scheme's design. What makes this notable is the speed. Prior to the AI model's involvement, cryptographers would have needed to manually analyze the scheme, write custom code, and test hypotheses. The AI compressed that entire workflow into less than three days.

The economics are just as important. $100,000 is trivial for a well-funded lab or a nation-state, and it is increasingly within reach of mid-tier cybercrime organizations. In the past, breaking a 64-bit security level would have required either a massive parallel computation or a sophisticated mathematical breakthrough. Now it can be done with one specialized AI model and a modest cloud budget.

HAWK and the post-quantum race

HAWK is one of several proposed digital-signature schemes designed to resist attacks from both classical and quantum computers. While the world's most widely used signatures, such as ECDSA and RSA, are still considered secure against non-quantum attackers, they are vulnerable to Shor's algorithm. A sufficiently powerful quantum computer could use that algorithm to derive a private key from a public key, enabling an attacker to drain wallets, forge transactions, or impersonate users.

To avoid that scenario, standardization bodies like the U.S. National Institute of Standards and Technology have been evaluating post-quantum algorithms for years. HAWK is among the candidates that have been submitted as possible replacements for the signatures used in TLS certificates, digital documents, and blockchain platforms. The promise of HAWK is that it offers compact signatures and fast verification, making it attractive for applications where bandwidth and latency matter.

But the AI-discovered flaw throws cold water on that promise. The scheme's smallest parameter set was already intended to provide a security level of 64 bits, which is low but acceptable for low-value, short-lived keys. After the attack, that level collapses to 38 bits. Larger parameter sets, which were meant to compensate for security loss, become far less attractive because the same structural weakness may scale with them. The result is that HAWK may not be viable as a post-quantum replacement in its current form.

Bitcoin's quantum exposure

Bitcoin currently uses ECDSA with the secp256k1 curve, and newer versions of the network also support Schnorr signatures. Neither ECDSA nor Schnorr is quantum-resistant. If a sufficiently large and stable quantum computer existed today, an attacker could take a public key from the blockchain, compute the corresponding private key, and spend coins. For older transaction outputs that use pay-to-public-key scripts, the public key is visible immediately. For pay-to-public-key-hash addresses, the public key is normally hidden until the first spend, which buys some time.

That nuance is cold comfort, however. Once an address owner signs a transaction, the public key is exposed to the world. A quantum attacker monitoring the mempool could in theory attempt to reverse the public key in a race with the legitimate network confirmation. This is one reason why a post-quantum transition for Bitcoin is not just about changing a few lines of code. It requires careful planning around address formats, key custody, and consensus rules.

The migration debate

Bitcoin developers have been discussing quantum-resistant upgrades for years, but there is no consensus on timing. Some argue that quantum computers are still years or decades away from being able to break secp256k1. Others point to rapid advances in error correction, qubit counts, and quantum algorithms as evidence that the timeline is shrinking. The new HAWK result does not directly prove anything about quantum progress, but it does prove that the broader cryptographic threat model is evolving faster than expected.

The attack also raises an uncomfortable question for Bitcoin's governance. Upgrading cryptography on a decentralized network requires a soft fork or hard fork, which means nodes, miners, and users must coordinate. Even after an upgrade is proposed, it would take months or years to deploy and require significant testing. If quantum computers arrive sooner than expected, Bitcoin may not have enough time to transition gracefully. Cryptocurrencies that have stronger centralized control, like some federated sidechains, could move faster, but Bitcoin's decentralization is exactly what makes its upgrade process slower and more deliberative.

Implications for the broader ecosystem

The HAWK attack is not just a niche story for cryptographers. It highlights a new reality: AI models are becoming powerful enough to assist in cryptanalysis in ways that were previously the domain of elite researchers. Anthropic's Claude Mythos Preview is a product of the current AI arms race, and its ability to find a serious flaw in a proposed cryptographic standard suggests that automated discovery will become more common. That is a double-edged sword. On one side, AI can help cryptographers find weaknesses before malicious actors do. On the other side, the same technology can be used by attackers to probe systems with far fewer resources than traditional manual analysis would require.

For blockchain networks, the implication is profound. Consensus protocols, wallet software, and smart contracts are built on assumptions about the cost and difficulty of breaking cryptographic primitives. If AI can reduce that difficulty by a factor of millions, or even billions, as in the HAWK case, then many of those assumptions need to be revisited. It also strengthens the case for cryptographic agility—designing systems so that algorithms can be swapped out quickly when a weakness is discovered.

What this means for Bitcoin holders

For ordinary Bitcoin holders, the immediate takeaway is that their funds are safe. Bitcoin and ether signatures are unaffected by the HAWK flaw, and neither network uses HAWK. However, the broader signal is that the cryptographic landscape is shifting under everyone's feet. The same kind of AI-assisted analysis that broke HAWK could eventually be applied to the algorithms that do secure Bitcoin, should a sufficiently powerful AI model find a structural weakness in ECDSA or Schnorr. No such weakness is public knowledge today, and the cryptographic community has been studying these algorithms for decades without finding a practical break. But the HAWK result is a reminder that confidence is not certainty.

There is also a risk management angle. Bitcoin's long-term security model depends on the continued integrity of its current signatures. If post-quantum risk is considered a real threat, then holders with significant amounts of bitcoin may need to think about coin custody practices, such as not reusing addresses, moving coins into schemes that obscure public keys until the last moment, and watching for proposed network upgrades. None of this is necessary at this exact moment, but the HAWK attack accelerates the conversation.

The clock is ticking

The HAWK episode is best understood as a stress test for the broader post-quantum migration. It demonstrates that the gap between a quantum computer and a classical AI-assisted attack is shrinking in a different way than expected. Many in the cryptography community believed that the main threat to current signatures would come from enormous quantum machines, which remain expensive and experimental. But now, classical attacks assisted by AI are improving so quickly that even proposed replacement algorithms are vulnerable. That makes the entire planning exercise more difficult, because the target is moving.

For Bitcoin, the path forward is not a single switch to one new algorithm. It will likely involve a suite of post-quantum signatures, careful analysis of trade-offs, and an upgrade process that can be executed over years. The HAWK scheme was supposed to be part of that future. Its weakening by an AI model in 60 hours is not just a technical footnote; it is a signal that the future is arriving faster, and with more unexpected tools, than the industry's original quantum planning suggested. The race to build quantum-resistant money is no longer only about quantum computers. It is also about the accelerating power of artificial intelligence.

Source:Coindesk News