AI Control Is Becoming a Business Issue
Artificial intelligence is often introduced to you as a simple productivity tool: write faster, answer customers, summarise documents and automate repetitive work. However, a major conversation at TechBBQ in Copenhagen has shifted the question. Instead of asking only what AI can do, founders, investors and operators are asking who controls the models, data and infrastructure behind it. TechCrunch reported on the TechBBQ discussion.
That question matters to your Malaysian SME because adopting AI also means depending on someone else’s platform. If a model changes its rules, becomes unavailable, produces a different answer or gains access to more business data than expected, your operations may be affected. You do not need to build your own AI model to take control. You do need to understand where your information goes, which decisions AI can make and what happens when the tool stops working.
What Happened at TechBBQ?
At TechBBQ, a technology conference in Copenhagen, discussions repeatedly returned to “sovereignty”: the ability of businesses, communities and countries to maintain control over the technology they rely on. The conversation gained urgency after Anthropic’s AI models Mythos and Fable reportedly became unavailable to users outside Europe earlier in 2026, disrupting at least one software team while other companies treated the incident as manageable. The reported incident and its business impact were covered by TechCrunch.
Ellen de Brever, an angel investor and head of partnerships at Novo Nordisk Foundation Cellerator, said the debate was moving away from machine capability and towards human judgment. The key issue was not simply whether AI could act independently, but what people were willing to give up. Her comments were reported by TechCrunch.
Privacy advocate Meredith Whittaker also warned that AI assistants integrated into operating systems could become a large-scale data collection mechanism. Meanwhile, Stability AI co-founder Emad Mostaque connected control of AI with broader questions about economic and political power. The event’s discussions also emphasised that face-to-face human relationships remain important even as businesses adopt more automated systems. TechCrunch reported these remarks and themes from the conference.
Why This Matters for Malaysian SMEs
For a Malaysian SME, AI sovereignty is not an abstract European policy debate. Imagine you use an AI assistant to draft quotations, a chatbot to answer customer questions and an automation platform to read invoices. If one provider changes its application programming interface, limits access in your region or suffers an outage, several everyday workflows may stop at the same time. Your company may not have a dedicated technology team to repair the problem quickly.
Data location and data handling are equally important. Your business may process customer names, telephone numbers, identification details, delivery addresses, medical information, employee records or supplier contracts. Sending this information to an AI tool without checking its terms, retention practices and security controls can create compliance and trust problems. Malaysia’s Personal Data Protection Department provides guidance and resources related to personal data protection through its official portal at pdp.gov.my.
Consider a Klang Valley accounting firm, a Penang manufacturer or a Sabah-based tourism operator. Each may use AI differently, but the control questions are similar:
| Business area | Control question | Practical safeguard |
|---|---|---|
| Customer service | Can the chatbot invent an answer or promise something your team cannot deliver? | Require human approval for complaints, refunds and unusual requests. |
| Sales and quotations | Can AI access confidential pricing, supplier terms or customer history? | Limit access to the documents and fields needed for the task. |
| Finance | Can the system approve payments or change records without review? | Keep payment release and bank changes under authorised human control. |
| Operations | What happens if the AI provider is unavailable? | Maintain a manual procedure and export important records regularly. |
Start With an AI Control Map
You can improve control without slowing every project. Begin by listing each AI tool your staff use, including tools adopted informally by employees. Record its purpose, the information it receives, the people who can access it and the business process affected if it fails.
Next, classify tasks by risk. Drafting a social media caption is usually less sensitive than reviewing an employment document. Generating product ideas is different from recommending a medical action or deciding whether to approve a customer’s credit. The more a task affects a person’s rights, safety, employment or access to services, the stronger your review process should be.
AI should assist your team, not quietly become the only point of failure in your business.
Set clear internal rules in plain language. Your staff should know which information must never be pasted into a public AI chatbot, when a manager must review an output and how to report an inaccurate or inappropriate result. Keep a human accountable for customer-facing communication, financial decisions and sensitive records.
Build Resilience Before You Need It
The TechBBQ discussion highlights a practical lesson: dependency becomes visible only when access changes. You should therefore avoid building a critical workflow around one provider without a fallback. This does not necessarily mean subscribing to several platforms. It may mean keeping a documented manual process, storing source records in a system you control and ensuring that another employee understands how the workflow operates.
Ask every technology provider several questions before connecting business data:
- Where is your data processed and stored?
- Is your information used to train or improve the provider’s models?
- How can you delete, export or correct your data?
- What happens to your workflow if the service is interrupted?
- Can you restrict access by employee, role or department?
- Does the provider notify you about major policy or service changes?
Keep a record of important AI-generated work. For example, save the original customer request, the AI output and the employee’s final approval for sensitive cases. This gives you a basic audit trail and makes it easier to identify recurring errors. It also reminds your team that fluent writing is not proof of accuracy.
The Bigger Picture
The sovereignty debate is likely to become more relevant as AI agents move from answering questions to taking actions. An agent could update a customer relationship management record, prepare a purchase order, schedule a delivery or send a follow-up message. Each step may appear small, but combined actions can affect your reputation and operations.
For you, the central question is not whether AI should be banned or accepted without limits. It is where authority should remain. Let AI handle repetitive preparation, classification and suggestions. Keep approval, accountability and sensitive judgment with people who understand your customers and business context.
Malaysia’s diverse market makes human oversight especially valuable. Customers may communicate in Bahasa Malaysia, English, Mandarin or Tamil, use local expressions and expect culturally appropriate service. An AI system can support your team, but your staff must still check whether a message is accurate, respectful and suitable for the customer.
As you plan your next automation project, measure success by more than speed. Ask whether your team can explain the workflow, recover when the tool fails and protect the information entrusted to your company. The businesses best prepared for the AI era will not be those that surrender every decision to an agent. They will be the ones that use automation deliberately while keeping control of data, process and responsibility.
Your Next Step
This week, choose one AI-enabled workflow in your company. Write down the data it uses, the decision it influences, the person responsible for checking it and the backup procedure if the tool becomes unavailable. That simple exercise turns AI from an invisible dependency into a managed business capability.
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 →