Portugal's Council of Ministers approved on September 4, in Lisbon, the decision to train the public artificial intelligence Amália using the archives of Portuguese media outlets. The announcement was made by the Minister of the Presidency, António Leitão Amaro, who revealed the beginning of formal negotiations with companies in the sector to make the project viable. The financial allocation for this new development phase is 1.8 million euros, fully funded by the Recovery and Resilience Plan.
The primary objective of this capacity-building phase is to equip the model with the ability to read, listen and see text, voice and images with precision. The Executive intends to finalize data and archive sharing agreements with journalistic companies, emphasizing that the ministry intends to set an example in the ethical and contractual relationship between the national generative technology system and the media sector, safeguarding intellectual property and the value of informative content.
In addition to negotiations with private and public press, the algorithm's training will benefit from the documentary collections of several state entities, including the historical repositories of the National Library of Portugal, the Portuguese Film Library and various museums under state supervision. The amount will serve to finance the computing power, data curation and computational architecture necessary to optimize processing in the European standard Portuguese language.
Amália represents the first major open foundational model produced in Portugal, presented in July 2026 after the joint work of over 60 researchers from national university consortia. The tool was designed to act as an assistant to citizens and public servants, synthesizing complex legal regulations, guiding form completion and streamlining bureaucratic processes, including supporting people with special needs in their contact with public services. With this journalistic and cultural capacity-building phase, the executive sets the goal of consolidating digital and linguistic sovereignty against private multinational platforms.
