Will AI break elliptic curve cryptography (ECDSA), as predicted by Ethereum Foundation researcher Justin Drake, by end of 2026?
Delphi’s forecast
- Yesyour pick0.4%$0.00
- No99.6%$1.00
The forecaster strongly favors one outcome, so we can’t open a market for this question. You can still create this market yourself.
Forecast reasoning
As of October 17, 2026, no AI system has demonstrated a practical attack breaking ECDSA, and Coinbase's cryptography lead called Drake's warning unsupported "FUD" with no evidence of weakened mathematical assumptions. Drake's worst-case "months" timeline was a conjecture, not a demonstrated capability, leaving only 76 days until the deadline with no breakthrough achieved despite OpenAI's October 6 mathematics release containing no ECDSA breaks.
Settlement
Resolution criteria: - "Yes": Wins if, strictly after 2026-10-08T17:49:36.275Z and at or before 2027-01-01T00:00:00.000Z, a credible, independently verified demonstration is published showing that an AI-assisted mathematical method can practically break ECDSA — defined as recovering a private key from an exposed public key using commercially available or equivalent hardware (e.g., a large GPU cluster) within approximately one week, consistent with Justin Drake's own stated definition of "actually breaking" ECDSA — and this demonstration is confirmed by multiple independent cryptographic researchers or institutions. - "No": Wins if no such independently verified practical break of ECDSA via AI-assisted methods is published and confirmed within the window. Clarifications: - "Breaking ECDSA" requires a demonstrated, reproducible private-key recovery attack — not a theoretical paper, a conjecture, a partial result, or an improvement in algorithmic complexity that does not yet reach practical feasibility on available hardware within the defined timeframe. - The window for qualifying events is strictly after 2026-10-08T17:49:36.275Z and at or before 2027-01-01T00:00:00.000Z.
Jan 1, 2027, 12:00 AM UTC