Cornell AI Alignment Club

Learning resources for AI safety

Curated material spanning introductions, technical papers, policy work, newsletters, and fellowships.

Non-Technical Introduction to AI Safety • Newsletters, Podcasts, and Blogs • Research Fellowships • Technical Papers • Policy Papers

Non-Technical Introduction to AI SafetyFor a high-level, non-technical overview of arguments for caution around advanced AI systems, start here.

Blogs and YouTube

AI Safety in the News

Newsletters, Podcasts, and BlogsStay up to date with the latest developments in AI safety, policy, and governance through these newsletters, podcasts, and blogs.

Newsletters and Blogs

Podcasts

Research FellowshipsResearch fellowships and programs for students and professionals interested in AI safety, governance, and policy research.

AI Safety and Governance Fellowships

  • SPAR FellowshipPart-time, remote research fellowship that connects rising talent with experts in AI safety, policy, or biosecurity for 3-month research projects.
  • Pivotal Research Fellowship9-week, in-person London fellowship focused on AI safety and governance research with mentorship, workshops, and stipend support.
  • RAND CAST Fellowship (formerly TASP)Develops new generations of policy analysts and implementors at the intersection of technology and security issues. Fellows receive mentorship from RAND policy experts.
  • LawAI Seasonal Research FellowshipsWinter and summer fellowships offering law students, professionals, and academics paid, cutting-edge AI law research with close mentorship from LawAI’s research staff.
  • GovAI Summer and Winter FellowshipsStructured program designed to help researchers transition to working on AI governance full-time.
  • ML Alignment and Theory Scholars (MATS)Independent research and educational seminar program connecting scholars with top mentors in AI alignment, governance, and security for a 12-week residential program.
  • IAPS FellowshipFully funded, 3-month program for professionals from varied backgrounds at the Institute for AI Policy and Strategy.
  • Vista AI Law and Policy FellowshipSponsors students and recent graduates for independent research with mentor guidance or as research assistants with law professors and AI policy experts.
  • UChicago Existential Risk Laboratory Summer Research Fellowship10-week, in-person program for undergraduate and graduate students to produce high-impact research on emerging threats from AI and other existential risks.
  • ERA Fellowship8 weeks of fully-funded AI safety research with weekly mentorship from expert researchers. Work on technical safety, governance, or technical AI governance projects.
  • Astra FellowshipFully funded, 3–6 month, in-person program at Constellation’s Berkeley research center for AI safety research.
  • Vitalik Buterin FellowshipsFunds PhD students and postdocs working on AI safety and/or US-China AI governance research, administered by the Future of Life Institute.
  • Foundation for American Innovation Conservative AI Policy Fellowship8-week, fully-funded, work-compatible program designed for conservative policy professionals.
  • PIBBSS Fellowship3-month interdisciplinary fellowship for researchers studying complex and intelligent behavior in natural and social systems, mathematics, philosophy, or engineering.

International Fellowships

  • LASR Labs (London AI Safety Research Labs)13-week, in-person London technical AI safety research fellowship where participants work in teams on publication-oriented projects.
  • EU Tech Policy FellowshipProgramme empowering ambitious graduates to launch European policy careers focused on emerging technology.
  • Talos FellowshipThree-part program to accelerate European AI policy careers: 8-week online fundamentals course, 7-day Brussels policymaking summit, and optional 4–6 month paid placement at leading EU policy organizations.
Technical PapersIntended for researchers considering a transition to AI safety and advanced undergraduates who want to start technical work.

Mechanistic Interpretability

Mechanistic interpretability studies trained neural networks by reverse engineering the algorithms encoded in weights and activations.

Eliciting Latent Knowledge and Hallucinations

AI Evaluations and Standards

Goal Misgeneralization and Specification Gaming

Emergent Abilities

Survey and General Reading

Policy PapersFor students and practitioners interested in public policy, law, governance, and economics approaches to reducing AI risk.

Overviews and Surveys

Licensing, Auditing and Standards

Misuse and Conflict

Structural Risk