Project Eleven CTO Warns AI Could Break Bitcoin (BTC) Crypto Before 2030 Quantum Threat

Project Eleven CTO Flags AI Cryptanalysis

Connor Deegan, chief technology officer of quantum-defense research firm Project Eleven, warned in comments published Friday that artificial intelligence may crack the cryptography protecting Bitcoin (BTC) and Ethereum before quantum computers ever manage it. His case is narrow but pointed: AI systems have already surfaced attack methods against ciphers that remain at the research stage, although no scheme actually deployed on either chain has been broken. The argument sits inside the broader Bitcoin security debate, and nothing here turns on the Bitcoin price; the subject is the mathematics beneath every transaction. Both chains secure ownership with elliptic curve cryptography: a spender holds a private key that authorizes asset movements, while a public key lets third parties confirm the signature is valid. The attack in question targets signatures rather than the hash-based Proof of Work behind mining, but signatures are what authorize every spend. Quantum hardware changes the theory, since Shor's algorithm, a specialized computation, is known to break elliptic curve schemes. The AI risk is different: if advanced mathematical reasoning finds computational shortcuts nobody has documented, private keys could fall without any quantum machine at all. Justin Drake, a researcher at the Ethereum Foundation, has put a timeline on the worst case. He has said elliptic curve signatures could be broken within months rather than years, and that a large cluster of GPUs, the processors behind AI workloads, could restore a private key from an exposed public key in about a week. His precaution, which he calls bunker mode, is to move funds in advance to fresh addresses whose public keys have never appeared on-chain. Ordinary addresses hide the public key behind a hash function, so a never-spent address leaves the key invisible; holders who simply HODL deny an attacker the starting point. Neither figure is urging an urgent exodus today; both frame this as preparation.

Anthropic's HAWK Result Cost $100,000

The research record behind the warning is already concrete. On July 28, Anthropic announced that its AI model Claude Mythos Preview had found weaknesses in HAWK, a lattice-based signature scheme designed to resist quantum attack. The company reported that about 60 hours of study cut the estimated computational cost of a private-key recovery attack on HAWK-256 from 2^64 to about 2^38, a theoretical load roughly 67 million times smaller. The effort carried an API bill of about $100,000. The caveats matter. The result targets one specific HAWK configuration, attacks on larger key settings remain impractical, and HAWK belongs to the lattice-cipher family, a different branch from the elliptic curve schemes on Bitcoin (BTC) and Ethereum. Nothing in the announcement means Bitcoin's deployed cryptography has been broken. Both points matter for the risk debate: a demonstrated AI attack exists, but only in a laboratory setting against a scheme Bitcoin does not use. The quantum timeline comes from Project Eleven's co-founder and chief executive, Alex Pruden: he puts a practical quantum attack as early as 2030, with the probability of one exceeding 50% around 2033 to 2035. AI progress is harder to forecast, which is why Deegan warns the flaw could surface through machine reasoning before any quantum machine is ready. Defensive standardization is moving too. The US National Institute of Standards and Technology standardized post-quantum signature schemes in 2024, built on lattice problems and hash functions. But lattice ciphers are the same family Anthropic just dented in HAWK, and Project Eleven's executives argue that no single scheme carries a lifetime guarantee; the priority is an architecture that can switch ciphers when a new weakness appears. Dissent exists inside the industry: Yehuda Lindell, head of cryptography at Coinbase, responds that years of elliptic curve use have produced no evidence AI has weakened it.

Exposed Public Keys and Bitcoin's Upgrade Path

Our reading: this is a preparedness problem, not an incident, anchored in Anthropic's July 28 disclosure and NIST's 2024 standard. One sentence in that record carries the most practical weight: a public key becomes visible the moment an address spends. That puts the theoretical attack surface on old balances: any Crypto Whale with a long transaction history, exchange cold wallets and a Strategic Bitcoin Reserve treasury hold keys whose public halves are already on-chain. Drake's bunker mode addresses exactly that group, while holders of never-spent addresses face no exposure today. The harder part is protocol-level: replacing the signature scheme on Bitcoin (BTC) requires a consensus upgrade, and a slow-moving governance process may not complete after a threat materializes. The claim also stands contested among the qualified: Lindell's rebuttal remains unanswered by any demonstrated break of the deployed curve, so the burden of proof sits with the researchers, and the real question is how far migration can be pulled forward.