December 31, 2026

Full rules

As written on Polymarket for “Will OpenAI or Anthropic announce recursive self-improvement by December 31, 2026?”.

1. §1This market will resolve to "Yes" if OpenAI or Anthropic, or an official representative of either company, announces that the company has achieved recursive self-improvement by 11:59 PM ET on the specified date.
2. §2Otherwise, this market will resolve to "No".
3. §3Recursive self-improvement means a capability, as described in the company's announcement, in which the company's AI autonomously conducts the research and development needed to materially improve the general capabilities of its most capable or near-frontier AI systems, or to produce a more capable successor, with the improved system able to drive further such improvements.
4. §4Humans may set high-level goals and provide resources, oversight, and approval, but must not supply the substantive research choices or the technical solution.
5. §5Improvements that are toy, narrow, sandboxed, proof-of-concept, or limited to prompts, skills, or benchmark scores will not qualify.
6. §6A qualifying announcement must describe the capability as currently achieved in an official capacity.
7. §7Statements that AI is assisting, accelerating, or writing code for the company's research, statements of future plans or expectations, speculation, interviewer characterizations, leaks, model outputs, and third-party claims will not qualify.