Independent research testing four leading AI platforms on real SME scenarios reveals a pattern policymakers cannot ignore.
The G7 endorsed AI as a transformative opportunity for small businesses in Kananaskis in June 2025. The 2026 OECD D4SME Survey confirmed growing adoption. What neither addresses is what actually happens when a small business owner in an emerging market sits down and uses a free AI tool with a real problem.
To find out, I tested ChatGPT, Claude, Gemini, and Qwen Chat on six SME scenarios: cash flow analysis, professional communication in a second language, cross-border regulatory compliance, fraud detection, employment law, and business process digitalisation. Same prompts, free tiers, no follow-ups — the experience an SME owner gets walking in cold.
The results were largely encouraging. All four platforms correctly identified a mobile money fraud pattern in Kenya and named the right reporting authority. Every platform drafted a professional email for a Sri Lankan garment exporter that would have taken significant time to produce unaided. Business process advice was consistently structured and practical.
But the failures matter more than the successes.
ChatGPT invented a salary figure that could have talked a restaurant owner out of a hire they could actually afford. Every platform — without exception — gave an Indian chili sauce exporter comprehensive EU labelling guidance while failing to mention that she cannot legally ship a single jar without first obtaining an FSSAI Central Licence. The documentation exists in English, is publicly available, and is clearly structured. The models did not surface it. This is not a data availability problem. It is a training prioritisation problem — and a critical one for emerging market SMEs whose regulatory obligations sit on the origin side of the transaction, not the destination side.
The tools are genuinely useful. The gaps are genuinely dangerous. That distinction is where development policy needs to focus.
The full article can be found here.
