Overview

• New low‑cost and open‑source models such as DeepSeek, Qwen, Kimi and Llama have eroded the recent dominance of a few AI giants, creating a more competitive global landscape.
• Cutting‑edge benchmark performance is no longer sufficient for real‑world reliability, pushing developers to build domain‑specific integrations and accept greater responsibility for outcomes.
• The push for AI sovereignty reflects security and economic goals, with the United States and China treating AI as a national priority and countries seeking soft power through widely adopted models.
• Mexico’s Universidad Tres Culturas has rolled out an AI ecosystem from high school through university, including faculty training, a master’s program, and modules that teach ethics, prompt design and data validation.
• Political experimentation continues to test AI’s role in representation, exemplified by the 2024 ‘IA Steve’ project in the UK that explored new interactions between voters and their representatives.