Resolves YES if public evidence available by December 31, 2027, 23:59 UTC confirms that OpenAI already possessed the following capability when this market was created on October 8, 2026:
A method that generally recovers a secret scalar from an independently generated, full-strength public point on secp256k1, NIST P-256, or Curve25519’s standard prime-order subgroup, within seven days using computing resources available to OpenAI at that time. Breaking any one of these curves is sufficient.
Confirmation must come from NIST or a comparable national cryptographic standards body, or at least two named cryptographers independent of OpenAI who directly examined the evidence. They must confirm both the capability and when OpenAI had it. A confidential audit or controlled challenge can qualify; public attack code or compromising anyone’s keys is not required.
Toy or specially weakened instances, implementation flaws, leaked keys, merely theoretical speedups, and attacks requiring future hardware do not count. Otherwise NO; this means no qualifying confirmation by the deadline, not proof that no secret result existed.
Background: Justin Drake warned about a possible future break, while Scott Aaronson reported that labs were investigating cryptographic attacks. Neither is confirmation of success.
1,000