Mark Miller discusses the connection between software engineering and economics in AI. He highlights the unity between large software systems and large societal systems, following Hayek. He suggests that studying the emergence of intelligence in human institutions within society to uncover useful parallels for AI institution design. Miller discusses James Madison’s approach to the alignment problem and contrasts it with the unipolar takeover perspective. He emphasizes building transparent, accountable, and incorruptible institutions through technology and voluntary rule-based frameworks. One case in point is object capabilities for computer security and its application to secure and decentralized interactions in AI ecosystems. Finally, he explores challenges and pathways towards secure foundations, including AI in software engineering, usable security, and informed consent.