According to Woofun AI, potential breakthroughs in artificial intelligence within mathematics are triggering deep anxiety among top cryptographers about the foundations of public-key encryption.
Johns Hopkins University cryptographer Matthew Green issued a stern warning that OpenAI's latest advances could render ineffective the core security mechanisms protecting the internet and cryptocurrencies. Although there is currently no conclusive evidence that encryption has been broken, this theoretical risk has forced the industry to re-examine its security assumptions.
The root of this concern lies in a remarkable capability disclosed by OpenAI on October 6: an undisclosed AI model generated 722 mathematical research reports, covering 372 sets of research results, and attempted to solve approximately 4,000 long-standing unresolved scientific problems. Data compiled by Woofun AI shows that while these results are still being evaluated, the pattern-recognition ability they demonstrate has alarmed experts.
Bitcoin and Ethereum currently rely on elliptic curve cryptography to secure transactions. The core of this technology lies in the computational near-impossibility of deriving a private key from a public key. However, if AI can discover mathematical shortcuts that humans have missed for decades, attackers could directly obtain private keys, putting Bitcoin wallets with exposed public keys and Ethereum accounts that have already performed digital signatures at risk of having their funds stolen.
A deeper concern is the blurring of security boundaries. Green pointed out that systems originally believed to have 128-bit security may actually only have 96-bit or 108-bit security. A decline of just 20 bits in security level can reduce attack costs by roughly one million times.
In addition, the post-quantum standards ML-KEM and ML-DSA approved by the U.S. National Institute of Standards and Technology in 2024 rely on lattice-based mathematical algorithms that could likewise become targets of AI attacks, forcing the industry to reassess the effectiveness of existing protections.
Faced with this potential threat, sharply divergent technical response paths have emerged within the industry. Blockstream CEO Adam Back said bluntly that such warnings are "unfounded" and presented his team's SHRINCS hash signature system based on the SHA-256 algorithm. The system has been verified on Bitcoin's Liquid sidechain, with a minimum signature length of only 324 bytes, demonstrating extremely high efficiency.
However, hash signature technology is not a universal key. It cannot directly replace public-key encryption and key exchange functions, and promoting it on Bitcoin's mainnet would still require a complex protocol upgrade process. The limitations of this technical alternative make a strategy of relying solely on algorithm upgrades to defend against AI threats appear inadequate, especially when AI may simultaneously weaken multiple mathematical foundations, making a single technical line of defense highly vulnerable to being breached.
Market participants have already taken action. Charles Edwards, founder of Capriole Investments and a member of the BeInCrypto market intelligence expert committee, said he has stopped using DeFi services and significantly reduced the number of wallets and accounts to avoid potential systemic risks. He warned that if mainstream encryption technology is broken by AI, it will trigger market chaos.
At present, there is no publicly verified AI attack capable of cracking the core mechanisms of Bitcoin or Ethereum, but the real crisis lies in the speed of the industry's response. If AI eventually finds a way to break digital signatures, online banking services, and secure internet communications, whether the entire Web3 ecosystem can complete a comprehensive upgrade of its security systems before a disaster occurs will be the ultimate test determining its survival.