AI for Science
Menu

Safety · 7 items

Safety is part of picking the tool.

The same tools that design a protein or speed up an analysis can design something harmful, leak data that should stay private, or score well on a test without doing the work. If you lead a lab or a science company, safety is part of picking tools: know what a model was tested on, keep sensitive data out of tools you have not checked, and know where to send people who want to work on the problem.

Roles

3 items
  • Future of Life Institute AI Industry Associate/Lead, EUR 82k-150k

    Future of Life Institute is hiring an AI Industry Associate or Lead, EUR 82k to 150k. Closes November 15 (per the post). Posted Oct 15, 2026 · Deadline Nov 15, 2026 Source on LinkedIn

  • Atlas Ignota (Convergent Research): 10 Founders-in-Residence and Field Strategists for AI infrastructure security, with Coefficient Giving

    Atlas Ignota, with Convergent Research and Coefficient Giving, is hiring 10 founders-in-residence and field strategists to secure AI infrastructure against nation-state threats. Posted Sep 8, 2026 Source on X

  • Oxford interpretability interns, RAs, postdocsClosed

    Oxford interns, research assistants and postdocs in automated interpretability: systems that explain and correct large models. Posted Nov 3, 2025 · Deadline Nov 15, 2025 Source on X

Programs

2 items
  • FAS Policy Entrepreneurship Fellowship AI Safety, $5k stipend, 5 hrs/weekClosed

    FAS Policy Entrepreneurship Fellowship on AI safety, 5 hours a week with a $5k stipend. For scientists who want to shape policy. Posted Sep 18, 2026 · Deadline Sep 7, 2026 Source on LinkedIn

  • Lisbon AI Safety Hub 8-week non-technical fellowshipClosed

    Lisbon AI Safety Hub: an 8-week fellowship for people without a technical background. Posted Sep 18, 2026 · Deadline Sep 30, 2026 Source on LinkedIn

Reading

2 items
  • Map of AI safety programs for newcomers

    A map of AI safety programs for newcomers. Send it to the person on your team who keeps asking where to start. Posted Sep 1, 2026 Source on X

  • Donor interviews: AI safety funders want demonstrated wins

    Interviews with AI safety donors: they want to see demonstrated wins. Useful if you plan to fund or apply for safety work. Posted Aug 19, 2026 Source on LinkedIn

Also in the hub

Closed items stay listed with a closed label so you can watch for the next round.

  • Biology benchmark updated to block agents gaming it

    Agents gamed a biology benchmark through web access, and scores fell after the fix. When you read a score, ask how the test was locked down. Posted Sep 23, 2026 Source on X

  • Anthropic Life Sciences Verification Program: access to Claude on biology work

    Anthropic's route to fuller access for verified biology work. Anthropic says it exists to prevent biological misuse. Posted Sep 18, 2026 Source on LinkedIn

  • AI used to create new viruses for first time

    Scientists used AI to create new viruses for the first time (The New York Times). Hope for medicine, and a clear dual-use risk. Posted Aug 6, 2026 Source on X