Policy research on advanced AI security and governance. www.iaps.ai
A policy memo proposing a national CAISI+ center in the US Commerce Department for frontier model evaluations of catastrophic risks, emergency assessment, and AI security research, with budget and infrastructure recommendations.
Proposes a whistleblower incentive program modeled on the SEC's to improve enforcement of U.S. export controls on AI chips and other compute-related technology.
Argues the US government must help AI developers secure unreleased internal models against theft and sabotage, since stolen frontier models could strengthen Chinese military and economic capabilities.
Argues that AI chip smuggling is widespread and that the US must strengthen its AI chip export controls, eliminating smuggling as a first step to outcompete China.
An IAPS linkpost arguing that AI companies' safety research leaves important gaps and that governments and philanthropists should fund work to fill them.
Argues that Southeast Asian countries should join AI governance and safety discussions by engaging the public and building local AI research capacity, citing contextualized safety approaches in the region.
Proposes technical mechanisms to track AI chips after export so that controls on dangerous AI capabilities can be targeted and enforced, rather than applied as blanket restrictions.
Examines the design of the international AI Summit series, including hosting, secretariat, participant selection, agenda and frequency, and proposes six recommendations to formalise it and strengthen its impact on frontier AI safety governance.
Compares US and Chinese policy approaches to governing general-purpose AI, identifying divergences in regulation focus, principles, and international governance, and their implications for cooperation on AI safety and security.
Argues that international scientific assessments of AI risks need a coordinated process among the UK-led report, the UN, the OECD and AI Safety Institutes, and discusses how these actors could work together.