Did DeepSeek receive unannounced assistance from OpenAI in the creation of their v3 model?
➕
Plus
21
Ṁ7930
2026
12%
chance

Resolves "Yes" if, at the time of resolution, it is known publicly that DeepSeek received some form of assistance from OpenAI employees in the creation of their v3 model (such as model weights, technical assistance, or etc.).

(The intent here is to estimate the odds of OAI surreptitiously supporting DeepSeek to create the impression that China is closer to the US in an AI race than would otherwise appear)

Clarifications:
1. It does need to be voluntary assistance: hacks would not count. If the hack is via a cybersecurity flaw so blindingly obvious that I could have exploited it (I have no training in cybersecurity), and I have reason to believe it was deliberate, resolves YES.

2. If unknown at close, resolves NO.

#OpenAI
#AI Regulation
#Q*
#OpenAI Crisis
Get Ṁ1,000 play money
4 Comments
Sort by:
filled a Ṁ50 NO at 1.0% order

Remember point 2 of the description!

Deepseek kind of feels like a sputnik moment

  1. Does it have to be voluntary? How does espionage count?

  2. Does this resolve NO if unknown at the close date?

@IsaacLiu It does have to be voluntary, and it will resolve NO if unknown at close, yes.

Related questions