Why Sovereign
Over 180 languages are spoken in the Philippines. Almost none of them are meaningfully represented in the world's frontier AI models. Budots Media's priority project is a sovereign Filipino language model — built in the Philippines, for Filipino languages, under Filipino control.
Frontier models handle Tagalog passably and Cebuano poorly. Waray, Hiligaynon, Bikol — barely at all. A nation of 110+ million people risks running its digital life on models that neither understand its languages nor answer to its institutions. Sovereignty here is practical, not political: education, government services, news and commerce all need AI that works in the languages people actually speak.
Our starting position is unusual for an AI lab: a decade of original, rights-cleared multimedia in Cebuano and Tagalog. Thousands of hours of interviews, voiceovers, subtitled films and news footage from our own productions — training data most labs cannot buy.
We spent ten years accidentally building the dataset. Now we are deliberately building the model.
The roadmap runs from a text model for Cebuano and Tagalog toward a multimodal generative model — speech, image and video — trained on our archive. The AI pivot began right after the release of ChatGPT, and is funded through commercial production work, growth markets in Arabia, India and SE Asia, and European Structural Funds via BUDOTS MEDIA POLAND.
← All stories