Why OpenAI’s disputed maths breakthrough matters to your business
A dramatic story is unfolding around OpenAI’s reported solution to the Navier-Stokes problem, a mathematical challenge that has remained unresolved for roughly 90 years. The issue is not only whether an artificial intelligence system can produce an important proof. It is also about how businesses should manage AI-generated work, protect user data, verify results and explain where ideas came from.
For a Malaysian SME, this matters even if your business has nothing to do with advanced mathematics. You may already use AI to draft quotations, classify customer enquiries, summarise meetings, create marketing copy, analyse sales records or automate follow-ups. As AI becomes more capable, the central business question is shifting from “Can the tool do this?” to “Can you trust, verify and govern what it produces?”
What Happened
OpenAI said it had found a solution to the Navier-Stokes problem, one of seven Millennium Prize Problems. According to The Verge, the company said its internal AI model, supported by 10,000 concurrent agents, produced the result. The problem concerns the behaviour of flowing liquids and gases and has remained unsolved for around nine decades.
The announcement quickly became controversial. One day earlier, New York University mathematician Tristan Buckmaster and Anthropic researcher Levent Alpöge had published findings on a related problem. Buckmaster said he had contacted OpenAI after learning that the company had heard about their progress. He raised questions about whether work created in Codex sessions could have influenced OpenAI’s result. The Verge reported that OpenAI said no specific user data was accessed, while also saying it could not rule out the possibility that de-identified data from product usage had helped improve its models.
OpenAI staff member Sebastien Bubeck said the company had not seen the researchers’ work before it became public and that the proofs and precise results were different. Buckmaster disputed that interpretation. OpenAI also said it did not intend to claim the $1 million Millennium Prize. The disagreement has not only created a debate about the mathematics; it has highlighted questions about data boundaries, attribution and the difficulty of proving what an AI system has or has not used.
Why This Matters for Malaysian SMEs
First, AI output should be treated as work requiring review, not as an unquestionable answer. Imagine using an AI assistant to prepare a supplier agreement in Bahasa Malaysia and English, generate a tax-related spreadsheet or recommend stock levels for a Klang Valley outlet. A fluent response can still contain an incorrect assumption, an outdated rule or a calculation error. You remain responsible for the document or decision sent to a customer, supplier, employee or regulator.
Second, you need clear rules about confidential information. A small Malaysian manufacturer may want to paste a production issue into an AI tool. A clinic may consider summarising patient-related information. A recruitment agency may upload candidate profiles. A property firm may share tenancy documents. Before doing so, you should understand the tool’s data settings, retention terms and access controls. The OpenAI dispute shows why the question “Will my information be used to improve the model?” deserves a direct answer rather than an assumption.
Third, attribution matters. If your marketing team asks AI to create an article based on a competitor’s website, the result may be too close to existing material. If your developer asks an AI coding assistant to generate software, the business should retain records of the prompts, reviews, tests and approvals. If your sales staff use AI to summarise customer calls, you should know which source recordings were used and whether the summary has been checked.
| AI risk | Example for your business | Practical control |
|---|---|---|
| Incorrect output | An AI-generated quotation contains the wrong quantity | Require human approval before sending |
| Confidentiality | Staff paste customer or supplier data into a public tool | Define approved tools and prohibited information |
| Unclear source | A blog post includes unsupported claims | Keep source links and verify every important statement |
| No accountability | Several staff edit an AI-created document | Record the reviewer and approval date |
Use AI to accelerate the work, but keep a person accountable for the final decision, document or customer promise.
How you can apply this now
Start by creating a simple AI register. List every tool used by your team, what it is used for, which employee owns it and what information may be entered. For example, a customer service tool may receive general product questions but not identity card numbers or full payment details. A writing assistant may work with a public product description but not an unreleased product roadmap.
Next, introduce a review level based on risk. Low-risk tasks such as brainstorming social media headlines may need a quick human check. Medium-risk tasks such as customer replies, inventory suggestions and internal reports should have a named reviewer. High-risk tasks involving legal commitments, employment decisions, health information, financial records or regulatory submissions should require specialist review before use.
Keep an evidence trail. Save the source documents, final version, reviewer name and date. For automated workflows, record what triggered the action and what approval was required. This is useful when a customer asks why a message was sent, when a staff member challenges a decision or when you need to improve a workflow after an error.
The Bigger Picture
The OpenAI story suggests that AI progress is moving into areas previously reserved for specialist researchers. That could eventually influence engineering, logistics, scientific analysis and product design. Malaysian SMEs may benefit from more capable tools that can reason through complex processes, identify patterns and coordinate several business tasks.
However, capability without transparency creates operational risk. An AI system may produce a valuable result while making it difficult to identify its sources, assumptions or route to the answer. In a small company, where one error can affect a key customer relationship, your advantage will not come from using the most impressive tool blindly. It will come from combining automation with clear permissions, careful verification and practical records.
The lesson is straightforward: experiment, but set boundaries before adoption spreads. Choose approved tools, limit sensitive data, verify important outputs and make responsibility visible. Whether AI is preparing a sales summary or attempting a legendary mathematical proof, trust must be built into the process rather than added after something goes wrong.
For your next step, select one repetitive workflow in your business and document its current process. Mark where AI may assist, where a human must review and what information must never leave your approved systems. That small exercise can make your automation safer, clearer and easier to scale.
Ready to Streamline Your Operations?
Technology moves fast. Your operations should keep up. AutoRunBiz builds AI systems that run your daily workflows — from WhatsApp order capture to accounting. Book a free 15-min ops audit →