Risk Reporting for Developers’ Internal AI Model Use
Proposes a harmonized standard for frontier developers' internal-use risk reports, aligned with California SB 53, New York RAISE Act, and the EU GPAI Code of Practice.
Research / Sources
IAPS
Policy research on advanced AI security and governance. www.iaps.ai
Research Report (25)Compute governance (23)Frontier Security (22)International governance & China (18)Policy and Standards (18)Erich Grunewald (16)Jam Kraprayoon (14)Link Post (14)Oliver Guest (11)Linkpost (10)Joe O'Brien (9)Chris Covino (8)Issue Brief (8)Oscar Delaney (7)Bill Anderson-Samways (6)Responses to RFIs/RFCs (6)Matthew Mittelsteadt (5)Renan Araujo (5)Theo Bearman (5)Blog Post (4)Cara Labrador (4)Shaun Ee (4)Asher Brass (3)International Strategy (3)Onni Aarne (3)Brianna Rosen (2)Clarissa Koh (2)Hamish Low (2)Karson Elmgren (2)Sambhav Maheshwari (2)Sarah Godek (2)Article (1)Cassia King (1)Commentary (1)Dave Banerjee (1)Jan Wehner (1)Request for Information (RFI) Response (1)Sumaya Nur Adan (1)
Focus All
Topic All
Date Any time
Crucial Considerations in ASI Deterrence
A IAPS memo analyses the MAIM deterrence argument, in which great powers would sabotage rivals' ASI races, reformulating its claims and assessing the evidence for and against each.
Policy Options for Preserving Chain of Thought Monitorability
A policy report arguing that competitive pressure could push AI developers away from human-readable chain-of-thought reasoning, undermining monitoring, and proposing governance options to preserve CoT monitorability.
The Hidden AI Frontier
Argues that frontier AI models kept hidden inside corporate labs pose dual-use and security risks, and proposes transparency and oversight measures for these privately held systems.
Managing Risks from Internal AI Systems
Argues that internal frontier AI systems face theft and sabotage risks plus internal rogue deployment, self-exfiltration and successor sabotage, and recommends security practices, expanded evaluations and government visibility.
Helping the AI Industry Secure Unreleased Models is a National Security Priority
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.
Mapping Technical Safety Research at AI Companies
Analyzes 80 safety papers from Anthropic, Google DeepMind and OpenAI (2022-2024), mapping where corporate safety research is concentrated, where gaps lie, and how incentives shape it.