Founder Advised Limiting AI Use in Journalism
Kevyn Wahuzue of Kanaky Tech warns that AI models lack the cultural accuracy required for Pacific journalism.
Updated on Sept. 28, 2026 in Artificial Intelligence
Live Poll
Do you trust artificial intelligence to accurately report news stories without human oversight?
Kevyn Wahuzue, founder of Kanaky Tech, has advised Pacific journalists to restrict their use of AI models to basic administrative tasks. This recommendation follows testing that revealed significant inaccuracies in how AI handles Pacific languages and cultural traditions.
Why it matters
AI models currently struggle to differentiate between distinct Pacific languages and often generate incorrect cultural descriptions, which can undermine the integrity of journalism. These errors stem from a lack of human judgment and established community relationships in standard model training.
Kanaky Tech developed Pasifika AI using a restricted set of 66 institutional sources to ground model responses. Every answer provided by the system includes cited references to these sources to maintain verifyability.
The players
Kevyn Wahuzue
Founder of Kanaky Tech who developed the Pasifika AI tool focused on institutional data sourcing.
Kanaky Tech
A technology company developing specialized AI models for Pacific language and cultural accuracy.
Te Hiku Media
A media organization based in Aotearoa New Zealand that develops specific speech technology for te reo Māori.
The details
Pasifika AI functions by limiting its training data to specific, vetted institutional sources to improve factual accuracy. During testing, researchers found that current AI models frequently confused Cook Islands Māori with New Zealand Māori and provided incorrect descriptions of the traditional Samoan pe'a tatau, a complex cultural tattoo. These failures highlight the difficulty AI faces in accurately capturing nuance without localized training, such as the speech technology efforts pioneered by Te Hiku Media for te reo Māori.
Timeline
September 28, 2026: Publication regarding AI use in Pacific journalism.
The Tech Race
This approach marks a departure from the industry-wide trend of training on massive, uncurated datasets toward a model of sovereign, institutional data curation. It highlights the competitive gap between broad-market LLMs and precision tools developed through local expertise and community partnerships.
Journalists and content creators in the Pacific region should prioritize administrative automation over editorial tasks when using current-generation AI tools. Users must verify all outputs against local, institutional sources to avoid repeating cultural errors found in non-specialized models.
The takeaway
The primary takeaway for professionals is that institutional grounding is currently the only reliable way to maintain cultural accuracy in language-based AI. Practitioners should monitor the development of localized, small-scale models as a more precise alternative to general-purpose language tools.
Further reading
For broader trends in model sourcing, explore Artificial Intelligence.
Source note: This article includes information reported by Pacific Islands News Association.
Live Poll
Do you trust artificial intelligence to accurately report news stories without human oversight?
