AI models put Bitcoin signature break risk at up to 76% by 2046
Bitcoin.com News asked four AI models to estimate the chance that Bitcoin's ECDSA and Schnorr signatures could be broken, after onchain analyst Willy Woo shared similar AI-generated figures.
Why it matters: The experiment highlights how far AI models disagree on when quantum computing or classical cryptanalysis could threaten the signatures that secure Bitcoin holdings.
Key points
- Grok put the 20-year risk at 60% and Claude at about 66%
- All four models named quantum computing as the main long-term threat, with two-year risks of 0% to 2%
Four AI models put the risk of Bitcoin's cryptographic signatures being broken at between 34% and 76% over 20 years, in an experiment run by Bitcoin.com News and published on 11 Oct 2026.
Bitcoin.com News then asked about four models to assess the same question over two, five, ten and 20 years, covering both quantum computing and classical or AI-assisted cryptanalysis.
The models diverged most at the longest horizon. Hangzhou Deepseek gave the most conservative 20-year figure at 34%, while OpenAI's ChatGPT-6 Astra put it at roughly 76%. xAI's Grok and Anthropic's Claude fell between them, at 60% and about 66%. All four saw far greater risk over two decades than in the near term, and all named quantum computing as the principal long-term concern.
In a follow-up, the models said near-term estimates stay low because practical attacks need fault-tolerant quantum hardware at scale, and no cryptographically relevant quantum computer exists. They noted that no practical break of secp256k1 is known and that AI advances alone do not amount to a cryptanalytic breakthrough.
Live updates on NewsFeeds:BitcoinAIDesk editor: Keshav Aggarwal (accountable for this page; this summary was not edited by hand)
How this was produced: an AI model wrote this summary of Bitcoin.com News’s report, which we credit and link above. Read our Editorial and AI policy. Spotted a mistake? See Corrections.
Not investment advice. Read the risk disclaimer.