Charles Hoskinson Links OpenAI Math Controversy to Midnight Privacy

Charles Hoskinson has connected the controversy surrounding OpenAI’s newly published mathematical work to Midnight’s privacy model. In a new video, he warned that researchers and entrepreneurs may expose confidential ideas when using centralized AI platforms.

OpenAI Proposes Navier–Stokes Solution

OpenAI has published an AI-generated solution to the Navier–Stokes existence and smoothness problem, one of the seven Millennium Prize Problems. The company also released a formal proof written in Lean, a system that uses software to verify mathematical reasoning.

The Navier–Stokes equations describe the movement of fluids, including water, air, gases and blood. The unresolved problem asks whether solutions in three dimensions always remain finite and smooth or can develop singularities under certain conditions.

OpenAI said the proposed solution was produced by a group of AI agents using a model more capable than GPT-6 Astra. The work still requires independent examination and has not been formally recognized as a completed Millennium Prize solution.

Hoskinson Questions the Origin of the AI Proof

The dispute intensified after mathematician Jay Cummings said he had worked on the proof for a year and documented his research through conversations with ChatGPT.

Hoskinson suggested that OpenAI’s system may have accessed Cummings’ work,

adopted his approach and then expanded, verified and formalized it through multiple AI agents. He compared the situation to a mathematician finding another researcher’s notes, completing the unfinished proof and publishing the result independently.

OpenAI has not confirmed that Cummings’ private conversations were used to produce the solution. Hoskinson’s account therefore represents his interpretation of the proof’s possible origin rather than a confirmed description of OpenAI’s research process.

He also placed the dispute inside a wider debate over scientific authorship. If an AI system develops an unfinished human idea into a formally verified proof, credit may need to account for the researcher, the model and the company operating the infrastructure.

Hoskinson Positions Midnight for Private AI Research

Hoskinson presented the case as an example of the problem Midnight was designed to address. Midnight is a privacy-focused blockchain associated with the Cardano ecosystem and built around programmable privacy, selective disclosure and zero-knowledge proofs.

According to Hoskinson, researchers and entrepreneurs should be able to use AI inside private environments where the model provider cannot inspect their conversations, research or business ideas. Such a structure would preserve access to AI inference while protecting intellectual property from the operator of a centralized platform.

Midnight uses NIGHT as its native asset, although Hoskinson’s argument concerned the network’s privacy architecture rather than NIGHT’s market role. He did not present a working application, technical design or deployment schedule for private AI research. Confidential execution would also require supporting computing infrastructure beyond the blockchain itself.

Hoskinson is extending Midnight’s proposed role from protecting financial and identity data to securing unpublished discoveries and commercial ideas during AI-assisted work. The Navier–Stokes dispute gives that proposal a specific target: keeping the contents and ownership history of sensitive AI sessions outside the control of the company providing the model.