Discuss the ethical and intellectual property challenges posed by AI companies using proprietary user data/platforms in scientific discovery, with reference to a recent controversy in mathematics.

The Navier–Stokes existence and smoothness problem is one of seven Millennium Prize Problems listed by the Clay Mathematics Institute, each carrying a US $1 million award [1]. OpenAI's September 2026 claim to have cracked a version of it has shown that AI's entry into frontier science raises sharper questions of consent, credit and ownership than of capability.

The controversy in brief

• On 8 September 2026, OpenAI claimed an unreleased internal model, run as roughly 10,000 concurrent agents over 88 hours, produced a Lean-verified finite-time blow-up proof for the forced 3D equations [2].
• Mathematicians Tristan Buckmaster (NYU) and Levent Alpöge (Anthropic), who had proved blow-up for the related 3D Euler equations weeks earlier, allege the model converged on their own narrow approach [3].
• The Clay Institute has not certified the claim; the problem remains officially unsolved [1].

Ethical challenges

• Consent in data use: OpenAI denies seeing the rivals' work but concedes de-identified Codex usage data may have informed training [2] — blurring the line between platform provider and competitor.
• Conflict of interest: a proposed "concurrent release" reportedly excluding an Anthropic-affiliated researcher makes attribution turn on employer, not contribution [3].
• Verification integrity: no independent mathematician had reviewed the roughly 100-page proof, letting a company announcement stand in for peer review [4].

Intellectual property challenges

• Ownership of research traces: whether ideas typed into a proprietary platform remain the user's IP is legally unsettled.
• Authorship of AI output: inventorship and authorship for AI-assisted work are still under deliberation at WIPO's Conversation on IP and Frontier Technologies [5].
• Purpose limitation: India's DPDP Act, 2023 offers a template — data collected for a service cannot be silently repurposed [6].

Scientific credit is ultimately a public trust, not a corporate asset. Enforceable purpose limitation on platform data, mandatory disclosure of AI assistance, and independent verification before publicity can let AI accelerate discovery while keeping attribution honest — the standard the Clay Institute's continued caution rightly upholds.