Leading artificial intelligence companies have started to use their own systems to accelerate research and development, with each generation of AI systems contributing to building the next generation. This report distills points of consensus and disagreement from our July 2025 expert workshop on how far the automation of AI R&D could go, laying bare crucial underlying assumptions and identifying what new evidence could shed light on the trajectory going forward.

CSET’s Andrew Lohn shared his expert perspective in an op-ed published by The National Interest. In the piece, he explains that AI-assisted hacking signals a deeper cybersecurity threat: not new tools, but the breakdown of core defenses like defense in depth against adaptive, large-scale attackers.

CSET’s Kyle Miller shared his expert analysis in an article published by WIRED. The article discusses how OpenAI’s new open-weight models are drawing significant interest from the U.S. military and defense contractors, who see potential for secure, offline, and customizable AI systems capable of supporting sensitive defense operations.

CSET’s Helen Toner was featured on the 80,000 Hours Podcast, where she discusses AI, national security, and geopolitics. Topics include China’s AI ambitions, military use of AI, global AI adoption, and recent tech leadership changes.

Red-teaming is a popular evaluation methodology for AI systems, but it is still severely lacking in theoretical grounding and technical best practices. This blog introduces the concept of threat modeling for AI red-teaming and explores the ways that software tools can support or hinder red teams. To do effective evaluations, red-team designers should ensure their tools fit with their threat model and their testers.

This report analyzes over 250 scientific publications that use open language models in ways that require access to model weights and derives a taxonomy of use cases that open weights enable. The authors identified a diverse range of seven open-weight use cases that allow researchers to investigate a wider scope of questions, explore more avenues of experimentation, and implement a larger set of techniques.

As AI agents become more autonomous and capable, organizations need new approaches to deploy them safely at scale. This explainer introduces the rapidly growing field of AI control, which offers practical techniques for organizations to get useful outputs from AI agents even when the AI agents attempt to misbehave.

Organizations looking to adopt artificial intelligence (AI) systems face the challenge of deciphering a myriad of voluntary standards and best practices—requiring time, resources, and expertise that many cannot afford. To address this problem, this report distills over 7,000 recommended practices from 52 reports into a single harmonized framework. Integrating new AI guidance with existing safety and security practices, this work provides a road map for organizations navigating the complex landscape of AI guidance.

Recent op-eds comparing the United States’ and China’s artificial intelligence (AI) programs fault the former for its focus on artificial general intelligence (AGI) while praising China for its success in applying AI throughout the whole of society. These op-eds overlook an important point: although China is outpacing the United States in diffusing AI across its society, China has by no means de-emphasized its state-sponsored pursuit of AGI.

CSET’s Jessica Ji shared her expert analysis in an interview published by Science News. The interview discusses the U.S. government’s new action plan to integrate artificial intelligence into federal operations and highlights the significant privacy, cybersecurity, and civil liberties risks of using AI tools on consolidated sensitive data, such as health, financial, and personal records.