OpenAI's AI Agents Tackle a Millennium Prize Problem: Who Should Get the $1M?

astralmeowmeow · reddit · 2026-09-11

Context: OpenAI reportedly used thousands of AI agents to attack a Millennium Prize Problem and says it won't claim the $1M prize even if the solution holds up. The author thinks that's the right call — but worries about next time.
Core question: companies can now deploy what amounts to a massive artificial research workforce against bounties designed to reward extraordinary human achievement. If humans direct the AI but thousands of agents do most of the search and reasoning, who deserves the prize?
Proposal: when a prize is solved primarily via large-scale AI research, the money should instead fund AI safety/alignment research, grants for human researchers, or the problem's field — because OpenAI can voluntarily decline $1M, but the next company might not.
The author invites discussion on where the line should be drawn.

More from AGI Musings