The AI tools you rely on daily are caught in a tug-of-war
Think about how your business uses AI today. Maybe you use ChatGPT to draft customer email replies, or an AI accounting tool to flag odd transactions. You probably don’t think about whether the model behind that tool is “open” or “closed.” But the outcome of that debate will shape what you can do, what it costs you, and how much control you have over your own workflows in the years ahead.
Right now, a handful of tech giants are pushing for tighter control over AI models. On the other side, some of the world’s most respected AI researchers argue that keeping AI open is essential to prevent a few companies from becoming gatekeepers. This isn’t academic. It directly affects which AI tools Malaysians get to use, how adaptable they are, and whether you’re locked into one vendor’s vision.
TL;DR: Open AI models let you customise, own, and run AI on your own terms. Closed models are convenient but lock you in. Three AI pioneers recently debated this and all agreed: openness matters for competition and for access. For Malaysian SMEs, the practical takeaway is to choose AI tools that let you keep control of your data and avoid single-vendor dependency.
What This Means
There’s a critical difference between “open source software” and “open-weight AI models.” Open source software shows you the code — like a recipe you can read and modify. Open-weight models release the trained “weights” — the mathematical settings that make an AI model work. You can’t easily inspect them like code, but you can run them yourself, fine-tune them, and host them on your own infrastructure.
Geoffrey Hinton, Nobel Prize winner, at the Ai4 conference in Las Vegas, warned that open-weight models are easier to abuse because “you train a big model and then you give people the weights.” But he also admitted it’s too late to turn back: “I think that battle’s been lost. We now have open-weight models, so the barrier to lots of people getting these big models, which was the cost of training foundation models, that barrier has disappeared.” Source
Open or closed isn’t an absolute choice though. Fei-Fei Li, CEO of World Labs, pointed out that “it’s very dangerous to make this a dichotomy between complete openness all the way to complete closedness.” She used nuclear physics as an analogy: scientific papers are published openly, uranium is regulated, and labs sit in between. Different layers can have different rules. Source
How This Applies to Malaysian SMEs
Let’s bring this home. You’ve probably adopted a cloud-based CRM, a payroll system, or an AI chatbot. Those tools may be running on closed models owned by companies in the US or China. That’s fine while everything works smooth. But what happens when the vendor changes pricing, changes their model, or stops supporting a feature you rely on? You lose control. The open AI movement gives you an alternative: run or customise models that you actually own.
Take data privacy. Malaysian SMEs deal with customer data, financial records, and HR information. If you send that data to a closed AI service, you’re trusting a foreign company with it. Open-weight models mean you can run an AI locally or on your own server in Malaysia. That helps you comply with the Personal Data Protection Act and gives your customers more confidence. Andrew Ng, co-founder of Coursera, warns against gatekeepers: “I don’t want there to be gatekeepers. That limits how all of us can access AI.” Source For a small business, that gatekeeper could arbitrarily change your access, your features, or your ability to export your own data.
There’s also a market angle. Ng highlighted that countries like China are pushing open-weight models aggressively, and these are gaining adoption across Asia and Africa. He said, “you can see the way China’s model has tremendous accomplishment with Africa, for example.” Source For Malaysian SMEs, that means you already have access to capable open models from multiple regions. You’re not stuck with one technology superpower’s approach. You can pick tools that are optimised for your language — Bahasa Melayu, Chinese dialects, or Tamil — and your local context.
Consider your long-term growth. If you build a unique business process around a closed AI, you’re building on rented land. An open model lets you fine-tune it on your own data, create a custom assistant for your shop, and even offer that as a service to other local businesses. That’s how you create value that competitors can’t easily copy. The three pioneers each had different concerns, but they all agreed AI should remain widely accessible. Hinton said: “What we want to do is develop AI in a direction that helps people, and regulation will help us do that. You can’t leave it to people like Elon Musk and Mark Zuckerberg to decide how AI should be done.” Source
“The biggest risk for a small business isn’t AI going rogue. It’s getting locked into a tool you don’t control, on data you can’t export, at a price you can’t predict.”
What the three pioneers actually said
| Researcher | Stance | Key concern |
|---|---|---|
| Geoffrey Hinton | Open weights are risky but unavoidable | Models can be fine-tuned for cyber attacks |
| Andrew Ng | Strongly pro-open | Gatekeepers will limit access and innovation |
| Fei-Fei Li | Nuanced openness | False debate; we need both open and closed layers |
Source for all quotes and stances
Practical takeaways for your business
- Audit your AI tools. Make a list of the AI services you use. Ask: can I export my data fully if I leave?
- Prefer tools that support open models. When a vendor says “open,” check if it means open-source code, open weights, or just an API.
- Keep sensitive data off shared closed systems. If you handle customer IDs or payroll data, consider running local or private AI.
- Test open alternatives. Try tools like Llama or other open-weight models for simple internal use cases — drafting, summarising, or classifying email.
- Think about the ecosystem. If a vendor controls the whole stack, you’re at their mercy when they decide what’s allowed.
The Bigger Picture
The open vs closed AI debate will define the next decade of business software. If closed models win, a few big companies control the pace of innovation, the features you get, and the prices you pay. If open models win, you’ll have more choice, more flexibility, and more ownership over your AI systems. Malaysia, as a market that imports technology, benefits from an open ecosystem. We get to adapt global AI to our context instead of accepting whatever one vendor offers.
You don’t need to become an AI policy expert. But you should start paying attention to whether your tools are built on open or closed foundations. Ask vendors where their AI runs. Ask about data ownership. Ask about export formats. The answers will tell you whether you’re building your business on soil you own — or renting from a landlord you can’t negotiate with.
Ready to Streamline Your Operations?
Your business should run itself. AutoRunBiz deploys AI agents to automate your daily operations — WhatsApp orders, invoicing, customer follow-ups, and accounting. Book a free 15-min ops audit to see where automation fits your business →