GenAI framework, Safety and Alignment R&D and AI for Public Good: invest in forward alignment (for example RLAIF) and backward alignment (evaluating trained models, testing for dangerous capabilities, mechanistic interpretability) with international cooperation among AI safety institutes; and pursue public good by democratising access with human-centric design and digital literacy (including against anthropomorphising AI and deepfakes), supporting SMEs, improving public service delivery, upskilling workers, and tracking and reducing the energy and carbon footprint of training and inference.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.