New events

• Closing the Democratic Loop: Automated Oversight for the AGI Era 16 June (online).
  Seminar run by Cooperative AI Foundation (CAIF) with Michiel Bakker (MIT) discussing how democratic accountability can work when institutions (especially frontier AI companies) are run by AI agents. Explores automated oversight systems for reading agent traces, testing commitments, and escalating violations.

• Aletheia’s Quest 29 June – 26 July (online).
  AI lie detection challenge by Cadenza Labs, NDIF, and Schmidt Sciences where teams build detectors for LLMs induced to lie, with white-box and black-box tracks. Teams with the best-performing and most innovative methods will share a $50,000 prize pool.

• Stop The AI Race – Protest on July 11 11 July (San Francisco, USA).
  Protest organized by Stop The AI Race calling on frontier AI company CEOs to publicly commit to pausing development if other major AI companies do the same. Will include rallies and speeches at the offices of OpenAI, Anthropic, and Google DeepMind.

New training programs

• GovAI U.S. AI Policy Program early September – early December (online or Washington D.C., USA). Apply by 28 June.
  12-week bipartisan part-time program by Centre for the Governance of AI (GovAI) for US policy professionals in government and think tanks. Involves guided discussions on general-purpose AI, expert Q&As, and a policy memo or op-ed.

• Beijing AI Safety Residency mid-September – mid-December (Beijing, China). Apply by 30 June.
  3-month self-directed residency at a new AI safety hub for researchers to live and work alongside Chinese counterparts, inspired by the Constellation residency. Aimed at helping manage frontier AI risks by improving international coordination with China.

New events

• EAGxBerkeley 2026 21–23 August (Berkeley, USA).
  Brings together people from the Bay Area and broader US West Coast EA communities to connect, exchange ideas, and find ways to have a larger impact – including in AI safety. There will be talks, workshops, meetups, and 1-on-1s.

• Intro to AI Evaluations 11 June (online).
  Free public workshop run by AI Safety Awareness Project (AISAP) exploring how modern AI systems work and the field of AI evaluations: the methods researchers use to measure what models are capable of, where they fail, and how they might become dangerous.

• AI Policy and Governance Overview 18 June (online).
  Another workshop by AISAP, aiming to help participants understand the policy conversation well enough to follow what is happening, evaluate proposals, and think clearly about what responsible governance might look like.

• Hackathon: Breaking Barriers to AI Safety 13 June (London, UK).
  Event run by LISA and BlueDot Impact helping people from primarily non-technical backgrounds to start work on AI safety. Participants work on group projects or portfolio development.

• ARC White-Box Estimation Challenge 18 June – 19 September (online).
  Contest by Alignment Research Center (ARC) and AIcrowd to improve estimation algorithms for random MLPs. Includes a total prize pool of at least $100,000, focused on white-box methods leveraging model internals.

New training programs

• Lens Academy: Superintelligence 101 15 June – 27 July (online). Enroll by 7 June.
  New part-time course where participants read and discuss the book “If Anyone Builds It, Everyone Dies” together with a group. Involves weekly sessions exploring the arguments, evidence, and implications with fellow learners and a volunteer navigator.

• Iliad Fellowship: Fall 2026 7 September – 4 December (London, UK). Apply by 22 July.
  Mentored 3-month research fellowship in applied mathematics for alignment, run by Iliad in collaboration with PrincInt. Aimed at PhD or postdoctoral (or equally experienced) researchers in math, physics, or a related field.

• Iliad Intensive: August 2026 10–28 August (Berkeley, USA). Apply by 22 June.
  Accelerated 3-week introduction to technical alignment, with an emphasis on broad coverage of the field. Aims to advance foundational research in AI safety by folding mathematical expertise into the field. Applicants are selected on the basis of estimated mathematical strength.

• Iliad Intensive: September 2026 7 September – 2 October (London, UK). Apply by 27 July.
  As above but runs for 4 weeks.

• Iliad Intensive: October 2026 5–30 October (Berkeley, USA).
  As above.

• Iliad Intensive: November 2026 2–27 November (London, UK).
  As above.

• ML4Good Technical: UK September 2026 2–10 September (Southeast England). Apply by 1 July.
  8-day intensive bootcamp on technical AI safety aimed at people pursuing a career in the field. Curriculum covers technical foundations, AI safety strategy, and technical governance – ending with a capstone project. All backgrounds are welcome.

• ML4Good Technical: Europe September 2026 26 September – 4 October (Lyon, France). Apply by 1 July.
  As above.

• ML4Good Governance: Europe September 2026 14–22 September (Lyon, France). Apply by 1 July.
  As above but covering AI governance, strategy, communication, and career planning.

• ML4Good Governance: Europe October 2026 19–27 October (Lyon, France). Apply by 1 July.
  As above.

• AI Security Bootcamp (AISB): Vegas 2026 2–8 August (Las Vegas, USA). Apply by 21 June.
  7-day intensive for security professionals working to secure frontier AI systems. Participants complete pre-work before the program to establish baseline ML fundamentals, followed by an immersive week delivered through demos, lectures, guest speakers, and hands-on exercises.

• Introduction to Cooperative AI: Summer 2026 6 July – 28 August (online). Apply by 12 June.
  8-week course by Cooperative AI Foundation (CAIF) which aims to deepen participants’ understanding of the field and help them prepare to contribute. Includes weekly exercises, group discussions facilitated by field practitioners, and a dedicated community.

New events

• EAGxBerlin 2026 13–15 November (Berlin, Germany).
  Effective altruism conference bringing together attendees from around Europe to explore how evidence and reasoning can guide efforts to do good – including in AI safety. There will be talks, workshops, panel discussions, and 1-1 meetings.

• EAGxIndia 2026 19–20 September (New Delhi, India).
  As above but for India and surrounding regions.

• EA Summit: Istanbul 2026 24 October (Istanbul, Turkey).
  As above but for 1 day only and in Turkey. Will be mostly in Turkish.

• Vegas AI Security Forum ‘26 6 August (Las Vegas, USA).
  ~200-person event bringing together researchers, engineers, and policymakers working to secure powerful AI systems. Includes keynote talks, workshops, fireside chats, panels, and tabletop exercises. Runs alongside DEF CON week.

• Physics-Informed Research for Ambitious AI Safety 10 June (online).
  Technical talk by Ari Brill (Principles of Intelligence) hosted by Safe AI Germany. Covers how methods from physics can help us understand deep learning systems, and how modeling the structure of natural data distributions could unlock deeper interpretability.

• Supercooperation: The Future of AI for Democracy 5 July (Seoul, South Korea).
  ICML satellite workshop co-hosted by Foresight Institute and Cooperative AI Foundation (CAIF), sponsored by Future of Life Institute (FLI). Explores AI tools for collective deliberation, power concentration risks under rapid AI progress, and democratic institutional resilience.

• Canada AI Safety – June Meetup 3 June (online).
  Monthly meetup supported by BlueDot Impact, AIGS Canada, and Trajectory Labs bringing together people curious about or experienced in AI safety to learn and connect. Includes lightning talks and open discussion.

• AI Safety Action Weeks May Edition 29 May (online).
  Hosted by Encode Africa for young African researchers and the diaspora. Explores the alignment problem, frontier labs’ approach to AI, and capability scaling – through keynotes, practical quests, and interactive learning-by-doing activities.

New training programs

None this week.

New events

• Governing AI in the Wild: an AI Policy Hackathon 13 June (Toronto, Canada).
  Hackathon run by BlueDot Impact at Trajectory Labs where participants craft policy briefs that govern AI after deployment. There will be cash prizes and a report of the winning briefs will be routed to institutions that need AI policy guidance most.

• EA Summit: Wellington 2026 19 September (Wellington, New Zealand).
  1-day effective altruism conference for people across New Zealand interested in using reason and evidence to do the most good possible. Includes talks, workshops, and networking on AI safety and other EA cause areas.

• EA Summit: Taiwan 2026 1 August (Taipei, Taiwan).
  As above but for Taiwan.

• Technical AI Safety Puzzle #1 14 May – 12 June (online).
  Competition by BlueDot Impact challenging participants to interpret a trained text classifier where 7 of 8 features sit on clean linear axes but 1 doesn’t. Contestants must identify the rogue feature, explain its representation, and train a model with a weirder encoding.

• If AI safety achieves one thing before 2027, what should it be? 28 May (London, UK).
  Pre-EAG London gathering at London Initiative for Safe AI (LISA) hosted by BlueDot Impact, asking what AI safety should achieve before 2027. There will be lightning talks from leaders in the field, followed by a mixer.

New training programs

• MATS: Autumn 2026 28 September – 4 December (Berkeley, USA or London, UK). Apply by 7 June.
  Fellowship connecting researchers with mentors in alignment, transparency, and security. Fellows conduct research for 10 weeks at offices in Berkeley or London, with the opportunity to apply for a 6–12 month funded extension.

• Human-Aligned AI Summer School: 2026 13–16 July (Prague, Czech Republic). Rolling applications.
  4-day intensive summer school focused on technical alignment research. Includes talks, workshops, and discussions covering current questions in alignment research and the broader framings that shape them.

• AI Policy Leaders Programme: Autumn 2026 10 September – 22 October (online & Brussels, Belgium). Apply by 14 June.
  Fellowship by Talos Network which trains aspiring policy professionals for careers shaping AI governance. Provides mentorship, training, and hands-on policy experience – including potential placement within a think tank or policy organization.

New events

• Global Challenges Project: Oxford June 2026 19–22 June (Oxford, UK). Apply by 21 May.
  3-day residential workshop by Kairos exploring foundational arguments around risks from advanced AI (and biotechnology). Includes guest speakers, career planning sessions, and 1-1s. Travel, accommodation, and meals provided.

• SPAR Demo Day: 2026 23 May (online). Apply by 21 May.
  Virtual showcase by Supervised Program for Alignment Research (SPAR), presenting 100+ technical and governance AI safety projects developed during their 3-month program. Includes poster sessions, lightning talks, and a career fair.

• Panel: Careers Behind the Scenes of AI Safety 22 May (online).
  Panel hosted by BlueDot Impact with field-builders and operations people from Apollo Research, Constellation, FAR.AI, and LISA discussing how they entered the field. Covers day-to-day work and what AI safety orgs are currently hiring for.

• AI Governance in Practise: Fireside Chat with Alan Chan & Cristian Trout 27 May (online).
  Moderated conversation hosted by Safe AI Germany (SAIGE) between Alan Chan (GovAI) and Cristian Trout (AI Underwriting Company) going beyond framework-level discussion to examine the mechanisms that govern AI systems.

• Getting an impactful job: a Probably Good workshop 15 May (online).
  Interactive 90-minute career session based on Probably Good’s How to Get a Job guide. Covers testable career hypotheses, networking, and applications, with hands-on exercises. Aimed at early- and mid-career people actively searching.

• Future-Proofing Europe: The AI Preparedness Report 2 June (Brussels, Belgium).
  Talk hosted by Brussels AGI Collective where the AI Preparedness Report’s leading author, David Janků, will walk through its findings, what it means for European AI policy, and the actions the authors argue should be prioritized now.

• The AI Safety Debate: Should we pause development of frontier models? 10 June (Chicago, USA).
  Inaugural Chicago meetup co-hosted by BlueDot Impact and Portal Innovations, featuring a debate on whether AI development should be paused. Aims to introduce attendees to AI safety. There are required pre-readings.

New training programs

• Impact Accelerator Program: Summer 2026 13 July – 23 August (online). Apply by 7 June.
  6-week program run by High Impact Professionals (HIP) helping experienced professionals transition into high-impact careers – including AI safety. Participants follow a guided workbook and attend weekly group sessions, and receive continued support after the program.

• Constellation Visiting Fellowship: Fall 2026 14 September – 18 December (Berkeley, USA). Apply by 12 June.
  Lightly-structured 3–6 month funded fellowship for full-time AI safety researchers to work from Constellation’s Berkeley research center. Fellows connect with other researchers and have their housing, travel, meals, and workspace covered.

• Hardware Assurance Programme 17–22 August (Cambridge, UK). Apply by 7 June.
  6-day technical program by Cambridge AI Safety Hub (CAISH) for engineers and researchers to build the mechanisms that let third parties verify what AI chips are computing, where they are operating, and at what scale. Includes talks and in-depth technical sessions.

• CISAC Fellowship: AI in Strategic Stability and Military Competition September 2026 – June 2027 (Stanford, USA). Apply by 8 June.
  Stanford Center for International Security and Cooperation (CISAC) academic-year residency for predoctoral, postdoctoral, and junior faculty scholars. Includes faculty mentorship, weekly research seminars, policy workshops, and simulations.

• CISAC Fellowship: Geopolitics of AI September 2026 – June 2027 (Stanford, USA). Apply by 8 June.
  As above.

New events

• Large-Scale AI Risks (LSAIR): 2026 23–24 June (Leuven, Belgium).
  Brings together researchers, policymakers, and practitioners from across disciplines to address the most pressing challenges of advanced AI systems – covering ethics, safety, and governance. Keynotes from Roman Yampolskiy, Laura Weidinger, and Werner Stengg.

• Secure Program Synthesis Hackathon 22–24 May (online).
  Sprint by Apart Research and Atlas Computing bringing researchers and engineers together for 3 days to prototype tools to verify AI-written code. Includes tracks on specification elicitation, validation, spec-driven development, and adversarial robustness.

• Compute & Control: The Physical & Policy Foundations of AI Governance 13 May (online).
  Joint talk by Naci Cankaya (Technical Governance Team at MIRI) on verification mechanisms making AI claims testable across distrustful borders, and Haakon Huynh (Columbia SIPA) on compute infrastructure and leverage over frontier AI. There will be an audience Q&A.

• EA Summit: Vietnam 2026 11 July (Ho Chi Minh City, Vietnam).
  Effective altruism conference for people exploring how evidence and reason can guide efforts to do good. Includes introductory sessions on EA concepts and high-impact areas like AI safety, plus skill-building workshops and meetups.

• EA Summit: Nigeria 2026 25 July (Abuja, Nigeria).
  Conference bringing together the Nigerian effective altruism community alongside experienced EAs from around the world to connect, exchange ideas, and explore actionable paths to impact – including in AI safety.

• AI Safety Hong Kong: Social #1 + Reading Group #1 14 May (Hong Kong).
  Combines discussion of ‘Machines of Loving Grace’ by Dario Amodei with social drinks and networking afterwards. First gathering hosted by AI Safety Hong Kong, welcoming all backgrounds with no prior engagement with AI safety required.

New training programs

• PRISM AI Safety Research Fellowship: 2026 7 June – 27 September (online). Apply by 25 May.
  4-month, part-time program producing peer-reviewed research in interpretability, governance, evaluation, and alignment. Research methodologies are selected by blind peer voting and refined through sprints, culminating in conference submissions or software releases.

• Effective Thesis Fellowship: Summer 2026 July–September 2026 (online). Apply by 3 June.
  Pairs students and early-career researchers with high-impact partner orgs to develop their thesis as a real-world research project. Students receive mentorship and guidance, and can attend group meetings and workshops with other fellows.

• BASE Fellowship: Fall 2026 8 September – 18 December (online). Applications open 22 May.
  13-week, part-time program by Black in AI Safety and Ethics (BASE) developing Black researchers in AI safety, security, and governance. Includes 5 weeks of foundational training followed by 8 weeks of mentored research across alignment, security, or governance tracks.

• EleutherAI Summer of Open AI Research 2026 13 July – 16 August (online). Apply by 15 May.
  Research program by EleutherAI letting people with little or no research experience learn by contributing to projects under mentor supervision. Hosted on the EleutherAI Discord.

New events

• MATS Spring Research Talks 20 May (London, UK).
  Evening of research presentations, Q&A, and discussion with current and past MATS fellows. Learn about what the research fellows are working on, build connections for new collaborations, and discuss current issues and paths forward in AI safety.

• LessWrong Community Weekend (LWCW): 2026 11–14 September (Berlin, Germany).
  Annual rationalist unconference which brings together aspiring rationalists from across Europe and beyond for 4 days of intellectual exploration and fun. Many activities are unrelated to AI safety, but those are optional and there are many AI safety people there to make it worthwhile.

• Third Symposium on AIT and ML: AI Safety Applications 27–29 July (Oxford, UK).
  Symposium on the intersection of algorithmic information theory (AIT) and machine learning, focused on AIT applications to AI safety theory and practice. Topics include Solomonoff induction, Kolmogorov complexity, embedded agency, and imprecise probability.

• One Bias After Another: Fixing Language Reward Models 6 May (online).
  Talk hosted by Safe AI Germany (SAIGE) where Daniel Fein and Max Lamparth (Stanford University) will explore why LLMs remain plagued by simple biases like tending towards excessively long/short responses, avoiding uncertainty, and agreeing with the user.

• AI Risk Content Hackathon 6 June (London, UK).
  1-day AI safety video creation sprint by BlueDot Impact, held at LISA. Creators will collaborate to produce at least one fully-edited short-form video about AI risk. There’ll be cash prizes.

• Where is AI in 2026 and Where is it Going? 12 May (online).
  Part 1 of a 3-part introductory series of workshops by AI Safety Awareness Project (AISAP) on AI’s current trajectory and what it means for all of us. Includes discussion of employment disruption, AI-enabled cyber threats, and human-AI relationships.

• The AGI Transition 14 May (online).
  Part 2 of the above series. Includes discussion of job automation, robotics, and AI replacing human connections.

• Superintelligent AI and Loss of Control 19 May (online).
  Part 3 of the above series. Includes discussion of misalignment risks, power and control dynamics, and policy decisions.

• Inside SaferAI: Making AI risk measurable 1 May (online).
  Talk by staff from SaferAI on how they build technical infrastructure that governments and companies use to understand risks posed by current AIs – from quantitative risk models to public ratings of frontier labs to the standards that codify EU AI Act compliance.

New training programs

• Arcadia Impact AI Governance Taskforce: Summer 2026 1 June – 21 August (online). Apply by 10 May.
  12-week part-time remote research program for experienced professionals transitioning into AI governance careers. Includes team-based policy research with partners from orgs like UK AISI and GovAI. Designed to be compatible with existing commitments.

• ARBOx4 (Alignment Research Bootcamp Oxford) 28 June – 10 July (Oxford, UK). Apply by 8 May.
  2-week bootcamp by Oxford AI Safety Initiative (OAISI) designed to rapidly build skills in AI safety. There will likely be both technical research and generalist streams. Applicants for the former should have basic familiarity with linear algebra, Python programming, and AI safety.

• plzdontkillus 1–31 July (Berkeley, USA). Apply by 15 June.
  Experimental creator bootcamp on AI safety where participants post short-form content every day – or go home. There’ll be workshops, collabs, talks, and mentorship from AI experts and content creators.

• The Secure Program Synthesis Fellowship 15 June – 30 September (online). Apply by 26 May.
  Program by Apart Research and Atlas Computing offering part-time research on mentor-led projects at the intersection of formal methods, AI systems, and security. Small teams tackle problems in specification, validation, and adversarial robustness.

• MAIA AI Safety Fundamentals Fellowship: Summer 2026 8 June – 2 August (Cambridge, USA). Apply by 22 May.
  Program by MIT AI Alignment (MAIA) on why AI safety matters and what’s being done about it – covering AI’s trajectory, misalignment, technical safety, policy, and careers in the field. Cohorts meet once a week in small groups.

• Technical AI Safety Project Sprint (May 2026) 18 May – 21 June (online). Apply by 10 May.
  Project-based course by BlueDot Impact for engineers and early researchers to work with an AI safety expert on a contribution to AI safety research or engineering. Includes mentorship, regular check-ins, and a published write-up. Runs ~monthly.

New events

• Summer Institute on Law and AI: 2026 15–18 July (Warrenton, USA).
  4-day gathering by the Institute for Law & AI combining law students, professionals, and academics to explore pressing issues at the intersection of AI, law, and policy. Includes presentations, fireside chats, and Q&As.

• Talent Needs in AI Safety – Tilman Räuker 28 April (Berlin, Germany).
  Talk hosted by Safe AI Germany (SAIGE) and Foresight Institute. Tilman Räuker, co-director of Pivotal Research, will break down the current talent needs across the AI safety ecosystem, what support researchers need, and lessons learned from running fellowships at Pivotal.

• How can I make AI go well? 30 April (London, UK).
  A mixer for professionals exploring a move into AI safety, hosted by the London Initiative for Safe AI (LISA). Aims to be a low-pressure evening to explore the field, understand the part you could play, and meet the people who are already in it.

• Colosseum: Auditing Collusion in Cooperative Multi-Agent Systems 23 April (online).
  Seminar by Cooperative AI Foundation (CAIF) as part of their ‘Updates in Cooperative AI’ series. Sahar Abdelnabi from the ELLIS Institute Tübingen will present a framework for auditing collusive behavior in LLM-based multi-agent systems.

• DC AI Security Forum 2026 18 June (Washington D.C., USA).
  Event bringing together people engaged in securing AI systems and infrastructure – including those from AI companies, U.S. government agencies, and research institutions. Includes workshops on securing model weights, AI hardware, and compute infrastructure.

• BlueDot Impact AI x Cyber Reading Group 24 April (online).
  Fortnightly reading group hosted by BlueDot Impact discussing recent papers at the intersection of AI and cybersecurity. Covers both security of model weights and how AI affects cybersecurity broadly.

New training programs

• Pivotal Research Fellowship: 2026 Q3 29 June – 28 August (London, UK). Apply by 3 May.
  Fully-funded, 9-week research program with a possible extension of up to 6 months. Includes mentorship, research management, workshops, talks, and workspace at LISA. Open to applicants from various backgrounds.

• CBAI AIxBiosecurity Summer Fellowship: 2026 8 June – 10 August (Cambridge, USA). Apply by 3 May.
  Fellowship by Cambridge Boston Alignment Initiative (CBAI). Fellows work with mentors and research managers, participate in weekly workshops and speaker events, and gain research experience and networking opportunities within the AI safety and biosecurity communities.

New events

• Minimal AI Safety Unconference (MAISU) 2026 24–27 April (online).
  Event for anyone who wants to help prevent AI-driven catastrophe. As an unconference, anyone can host a session on any AI safety-relevant topic. There will also be lightning talk sessions where teams from the current AI Safety Camp present their projects.

• ILIAD 2026 3–7 August (Berkeley, USA).
  Free 5-day unconference bringing together researchers to build scientific foundations for AI alignment. Includes participant-led sessions on singular learning theory, agent foundations, causal incentives, and scalable oversight.

• Inside Apollo: Science of Scheming & AGI Safety Products 30 April (online).
  Event hosted by BlueDot Impact featuring talks by Apollo Research’s ​​Marius Hobbhahn, Alexander Meinke, and Joping Chai. They’ll discuss their Science of Scheming research agenda, AGI safety evaluation products, and job opportunities at Apollo.

• Cambridge Conference on Catastrophic Risk (CCCR): 2026 17–18 September (Cambridge, UK).
  Event hosted by the Centre for the Study of Existential Risk (CSER) that aims to bring together the existential risk community, with a focus this year on the governance challenges shaping today’s world – including around AI safety.

• Global South AIS Hackathon 19–21 June (online).
  Hackathon by Apart Research bringing together researchers, engineers, and policy professionals from Latin America, Africa, and Asia to produce AI safety tools, evaluations, and policy research. Includes prizes and potential invitations to the Apart Fellowship.

• EAGxAustralasia 2026 27–29 November (Melbourne, Australia).
  Effective Altruism conference bringing together 300+ people working in various EA cause areas, including AI safety. There will be talks, workshops, discussions, and networking opportunities.

• EAGxSingapore 2026 29–30 August (Singapore).
  As above but in Singapore.

• IAS X BlueDot Impact – AI Safety Madrid 28 April (Madrid, Spain).
  Spanish-language meetup co-organized by IA Safety (IAS) and BlueDot Impact. There’ll be talks, open debate, and opportunities to connect with others. Open to all, whether you’re an expert or just getting started.

New training programs

• Iliad Intensive: June 2026 6 June – 3 July (London, UK). Apply by 22 April.
  4-week, full-time foundational alignment program focused on math-heavy alignment research over coding. There are 20 modules across alignment, learning theory, interpretability, agency, and safety guarantees. Includes a stipend, office space, and meals.

• MARS V 13 July – October 2026 (online or Cambridge, UK). Apply by 3 May.
  Part-time research program by Cambridge AI Safety Hub (CAISH) pairing teams of 2–4 with mentors to produce AI safety research across technical alignment and governance. Includes a 1-week kick-off sprint, compute credits, and research management support.

Notes

• This program is now also available online and the deadline has been extended:
  AFFINE Superintelligence Alignment Seminar 28 April – 28 May (online). Apply by 24 April.
  Month-long program providing promising newcomers to AI alignment a deep understanding of some large pieces of the problem, making them better equipped to work on mitigating AI existential risk. Top candidates may be offered a full-year fellowship.

New events

• AI Transparency Conference (AITC) 2026 5–6 June (Nuremberg, Germany).
  Research conference by COAI Research focused on advancing transparent and human-compatible AI systems. Covers interpretability, agent behavior, control, safety, and governance. Aimed at PhD students, postdocs, faculty, and research institutes.

• Pause Plan-a-Thon 2–3 May (online).
  Hackathon organized by AI Plans bringing together economists, policy folks, supply chain experts, and AI safety researchers to figure out what it would take to halt frontier AI globally. Judges include Joe Rogero (MIRI) and Yann Rivière (Epoch AI/Metaculus).

• Preparing for AGI 28 April (online).
  Inaugural webinar hosted by AI Safety Hong Kong, where Tzu Kit Chan (Atlas Computing) will discuss risks, governance gaps, and what alignment and accountability may look like in practice. No technical background required.

New training programs

• Dovetail Research Fellowship 2026 at latest 1 June – 10 August (online). Apply by 17 May.
  ARIA-funded 10-week mathematical AI safety research fellowship for undergrads through post-docs. Some fellows may work together, while others solo. Involves weekly group meetings, 1-on-1 check-ins, and an end-of-term write-up and talk. Applicants must be UK-based.

• Astra Fellowship: Summer 2026 14 September 2026 – 5 February 2027 (Berkeley, USA). Apply by 3 May.
  Fully funded 5-month fellowship by Constellation pairing emerging talent with senior advisors on AI safety research, governance, and strategy projects. Prior AI safety experience not required. Includes mentorship, career support, compute, and a monthly stipend.

• Generator Residency 15 June – 28 August (Berkeley, USA). Apply by 27 April.
  3-month program by Kairos and Constellation for AI safety generalists to pitch, build, and ship projects that build capacity and infrastructure across the AI safety ecosystem, and then get support landing full-time roles at orgs that need them.

• OpenAI Safety Fellowship 14 September 2026 – 5 February 2027 (San Francisco, USA). Apply by 3 May.
  6-month, full-time research fellowship focused on empirical AI safety. Fellows pursue research in areas of interest to OpenAI, including safety evaluation, robustness, scalable mitigations, and agentic oversight. Program includes mentorship, compute, and stipends.

• AIxBiosecurity Research Fellowship 6 July – 11 September (Cambridge, UK). Apply by 27 April.
  10-week, fully funded research fellowship run by ERA and Cambridge Biosecurity Hub. Fellows will work on projects mitigating biosecurity risks, including those amplified by frontier AI. Potential research directions include evals, red-teaming, and unlearning hazardous knowledge.

New events

• How To Start An Impactful Career In AI Governance 12 April (online).
  Talk by Carolin Basilowski from Talos Network covering concrete AI governance career paths, practical entry routes for newcomers, and low-cost opportunities to test your fit for policy work. Emphasis is on the EU.

• Australian AI Safety Forum 2026 7–8 July (Sydney, Australia).
  2-day interdisciplinary conference organized by Australians for AI Safety, aimed at bringing people together from across research, government, industry, and civil society to help shape Australia’s role in AI safety. There will be presentations, workshops, and structured discussions.

• AI Safety New Zealand Conference 2026 4–5 July (Christchurch, New Zealand).
  AI Safety Australia & New Zealand (AISANZ) is hosting its first AI safety conference in New Zealand. It aims to help discuss and build the frameworks – technical, ethical, regulatory, and otherwise – needed to ensure transformative AI is beneficial.

• 1st IJCAI Workshop on Safe Physical AI 15–17 August (Bremen, Germany).
  Workshop at IJCAI/ECAI 2026 exploring safety challenges in physical AI systems, from near-term autonomous robotics risks to longer-horizon threats from physically capable AGI. Includes talks, poster sessions, panel discussions, and a mentoring program for early-career researchers.

• Threat Models & Mitigations from Physical AGI 15 April (online).
  Talk and interactive Q&A hosted by Safe AI Germany (SAIGE) exploring how robotics safety paradigms scale when applied to AGI, including threat models like infrastructure capture, autonomous violence, and embodied manipulation.

• Secure & Sovereign AI Workshop 18–19 July (Berlin, Germany).
  Event hosted by Foresight Institute that brings together researchers, engineers, and funders to explore how to make AI an engine of defense in a multipolar human-AI world. There will be 3 interconnected focus domains: secure AI, private AI, and decentralized & cooperative AI.

• AI Immune System: Detection Challenge 1 April – 31 May (online).
  Competition by the AI Alignment Network challenging participants to build ML models that identify harmful or unsafe statements embedded within AI agent conversations, where dangerous intent is obscured by natural-sounding language. $3,000 prize pool.

New training programs

• CBAI Summer Research Fellowship in AI Safety ‘26 8 June – 10 August (Cambridge, USA). Apply by 12 April.
  Fully funded 9-week research fellowship by the Cambridge Boston Alignment Initiative (CBAI) covering interpretability, multi-agent safety, formal verification, and AI governance. Includes $10,000 stipend, housing, mentorship, and up to $10,000 in compute.

New events

• Vision Weekend United Kingdom 2026 5–7 June (London, UK).
  3-day conference from Foresight Institute aimed at researchers and builders across frontier science and technology. Includes an AI safety track led by Apollo Research covering emerging AI paradigms, and presentations from AI safety researchers.

• TAIGR 2026 10–11 July (Seoul, South Korea).
  Workshop at ICML 2026 bringing together ML researchers and policymakers to advance technical tools for AI governance – including assessment, verification, security, and ecosystem monitoring of AI systems.

New training programs

• ERA Fellowship: Summer 2026 6 July – 11 September (Cambridge, UK). Apply by 12 April.
  10-week, fully funded program hosted by ERA for researchers and entrepreneurs working on mitigating risks from frontier AI. Includes mentorship, stipends, meals, transport, visas, lodging, and events. Covers technical AI safety and governance research.

New events

• Lie Detection Competition 9 March – late summer 2026 (online).
  Competition run by Schmidt Sciences, NDIF, and Cadenza Labs advancing lie detection in LLMs. Red teams create datasets of on-policy LLM lies while blue teams build detection methods. Selected teams receive $10,000–$25,000.

• Technical AI Safety Demo Day 27 March (online).
  Showcase of top projects from BlueDot Impact’s latest Technical AI Safety Project Sprint. Includes lightning talks where each presenter gives a high-level overview of their project and findings, followed by breakout groups for deeper discussion of each project.

New training programs

• CAMBRIA: May 2026 18 May – 5 June (Cambridge, United States). Apply by 30 March.
  Full-time ML upskilling program run by Cambridge Boston Alignment Initiative (CBAI) for aspiring technical AI safety researchers. Based on the ARENA curriculum; covers transformers, mechanistic interpretability, and alignment science. Meals and travel support provided.

• CAMBRIA: July 2026 6–24 July (New York City, United States).
  As above.

• CAMBRIA: August 2026 10–28 August (Cambridge, United States).
  As above.

• TAISI Summer Intensive 2026 May–August 2026 (Toronto, Canada). Apply by 5 April.
  Bootcamp running 1 day per week for 1 month, organized by the Toronto AI Safety Student Initiative (TAISI) in partnership with Trajectory Labs. The aim is to train students to become impactful AI safety researchers. There will be 4 cohorts throughout the summer.

New events

• 10th Annual CHAI Workshop 4–7 June (Pacific Grove, USA).
  Event by the Center for Human-Compatible AI (CHAI) inviting researchers, practitioners, and policymakers working at the forefront of AI and AI safety to gather in a retreat-style setting to examine emerging research questions and shape progress in the field.

• MATS Winter 2026 Research Symposium 28 March (Berkeley, USA).
  Symposium celebrating the end of the Winter 2026 MATS cohort. Includes spotlight talks and poster sessions where scholars present their research. There will be an end-of-program party afterwards.

• Stop The AI Race Protest 21 March (San Francisco, USA).
  Peaceful protest calling on major AI company CEOs to publicly commit to a conditional pause on frontier AI development. There will be a speech by David Krueger, followed by marches to AI company headquarters.

• AI Safety in Austin – Happy Hour 13 March (Austin, USA).
  Inaugural meetup hosted by EquiStamp bringing together researchers, engineers, founders, and curious minds who care about getting AI right. Includes a short talk on EquiStamp’s research and work, plus food and drinks.

• SAIGE: AI Middle Powers Discussion Round 18 March (online).
  High-level discussion involving experts from AI policy and strategy. Topics will include strategies for European state capacity, the reality of brain drain, and the geopolitical impact of middle powers’ regulatory frameworks.

• SERI Symposium 2026: Emerging Technologies & Existential Risk 3 April (Stanford, USA).
  Event exploring the risks from emerging technologies, with particular attention on governance and technological solutions. There will be experts and researchers from across the Bay Area who specialize in a range of global risks, including that from AI.

• AI Safety Hub Nigeria: Fireside Chat 14 March (online).
  Inaugural fireside chat hosted by AI Safety Hub Nigeria unpacking the India AI Impact Summit and discussing what it means for Africa and developing economies in shaping global AI safety governance.

• AI Safety Fieldbuilding Hackathon 28 March (London, UK).
  Hackathon focused on building AI tools to improve the operational infrastructure behind AI safety field-building e.g. courses, funding programs, local groups etc. No prior field-building or AI coding experience required.

New training programs

• Research Accelerator Week 18–23 May (Cambridge, UK). Apply by 5 April.
  1-week program by Meridian Cambridge to help participants find collaborators and secure funding for independent AI safety research. Essentially a funding round with built-in support, it includes a dedicated funding pathway, expert guidance, collaborator matching, and fiscal sponsorship.

• AI Policy Strategy Fellowship: 2026 week of 25 May – 20 September (online).
  4-month fellowship by Successif equipping professionals with skills and frameworks to translate their research into effective policy action. Involves workshops, op-ed writing, and a capstone project.

• Talos Fellowship: Autumn 2026 12 August 2026 – 31 March 2027 (online & Brussels, Belgium). Apply by 27 April.
  Fellowship launching and accelerating European AI governance policy careers. Includes 8-week online AI policy fundamentals program, 7-day policymaking summit in Brussels, and optional paid 4–6 month placement at a leading governance organization.

• Dilemmas and Dangers in AI: Summer 2026 12 July – 16 August (online). Apply by 1 May.
  Fellowship run by Leaf for students aged 15–19 exploring how to steer AI toward benefitting humanity. Involves discussion groups, talks and Q&As with AI safety researchers, mentorship, and ongoing support. Weekly time commitment of 4–5 hours.

• AI and Society Fellowship 1 June – 21 August (San Francisco, USA). Apply by 24 March.
  Fully funded 3-month research fellowship by the Center for AI Safety (CAIS) for scholars in economics, law, international relations, and adjacent fields to research how advanced AI reshapes governance, policy, and society. Includes guest speakers and a final workshop.

New events

• Frankfurt AI Safety Meetup 19 March (Frankfurt, Germany).
  First event run by Frankfurt AI Safety, a new community for AI safety professionals and those exploring the field. There will be fortnightly meetups for hosting discussions and sharing research opportunities.

• EA Summit: Nairobi 2026 11 April (Nairobi, Kenya).
  Conference aimed at the Kenyan effective altruism community – including those interested in/working in AI safety. There will be talks, skill-building workshops, and networking. Kenya-based attendees will be prioritized.

New training programs

• Iliad Intensive: April 2026 4 April – 2 May (London, UK).
  Month-long introduction to technical AI alignment. Topics include the theory of RL, learning theory, interpretability, agent foundations, and scalable oversight and debate. Outstanding participants may be invited to the Iliad Fellowship. Application deadline 6th March.

• Second Look Fellowship: Summer 2026 15 June – 22 August (Chicago, USA).
  10-week research program run by the University of Chicago’s Existential Risk Laboratory (XLab). The focus is on producing high-quality replications of influential AI safety research, with fellows working closely with program leadership and peers.

• XLab Summer Research Fellowship: 2026 15 June – 22 August (Chicago, USA).
  Another 10-week in-person research fellowship by XLab, but this one is for students and early-career researchers working on AI safety, AI security, or nuclear security. Includes a stipend, housing, mentorship, and compute credits.

• Alignment Research Engineer Accelerator (ARENA) 8.0 25 May – 26 June (London, UK).
  4–5 week ML bootcamp with a focus on AI safety. The mission is to provide talented individuals with the skills, tools, and environment necessary for upskilling in ML engineering, for the purpose of contributing directly to AI alignment in technical roles.

• LASR Labs: Summer 2026 20 July – 16 October (London, UK).
  13-week full-time AI safety research program where participants are matched into teams of 3–4 and work with a supervisor to write an academic-style paper, with support and management from LASR.

• Cooperative AI Summer School 2026 3–7 August (Canada).
  5-day advanced program for early-career professionals already working in or studying cooperative AI. Participants attend talks from leading researchers and work on collaborative projects tackling real cooperative AI challenges.

• CLR Summer Research Fellowship 2026 29 June – 21 August (online or London, UK).
  8-week program by the Center on Long-Term Risk (CLR) to explore strategies for reducing suffering in the long-term future (s-risks) and work on related technical AI safety ideas. Fellows will be part of CLR’s team while working on their own research project.

• AI Security Bootcamp: Singapore 2026 20–26 April (Singapore).
  Intensive 1-week bootcamp run in partnership with Singapore AI Safety Hub (SASH) training security professionals to secure frontier AI systems. Includes demos, lectures, guest speakers, and hands-on red/blue team exercises.

New events

• AI Control Hackathon 2026 20–22 March (online).
  This Apart sprint, co-organized with Redwood Research and UK AISI, is aimed at developing more robust techniques to ensure AI systems are safe. Tracks cover control protocol design, ControlArena challenges, and red teaming & vulnerability research.

• AI Safety Founders Co-Founder Matching 25 February (San Francisco, USA).
  Networking meetup hosted by AI Safety Founders and sponsored by Seldon Lab, Apart Research, and BlueDot Impact. Aims to connect potential co-founders building organizations in AI safety – including startups, non-profits, and research orgs.

• What will mobilise the public to act on AI? 11 March (online).
  Talk by Social Change Lab sharing findings from a new study on which AI risks – ranging from job displacement to extinction – are most likely to motivate public action. Includes discussion with others working on AI safety public engagement.

New training programs

• Andrew Carnegie AI-Nuclear Policy Accelerator March–December (San Francisco, USA).
  10-month program by the Institute for Security and Technology (IST) giving nuclear policy professionals hands-on AI technical literacy through an in-person bootcamp, monthly virtual seminars, and applied policy labs in order to improve decision-making on AI-nuclear risks.

• Technical AI Safety Evals Course 2026 14 March – 21 June (online or Moscow, Russia).
  10-week AI safety course run by Monoid covering LLM and agent evals – including benchmarks, eval tools, and custom eval design. It’s in Russian and involves 5 weeks of learning and 5 weeks of mentored project work.

New events

• Technical Innovations for AI Policy (TIAP) Conference 2026 30–31 March (Washington D.C., USA).
  Second iteration of this FAR.AI conference bringing together policymakers and technical experts. Sessions include AGI preparedness and control, CBRN risks, export controls, hardware verification, and translating AI safety research into policy.

• Federal AI Regulation and Bipartisanship 21 February (online).
  PauseAI US webinar discussing current bipartisan support for AI regulation and how to avoid partisanship in AI safety advocacy, instead building a coalition of Americans across party lines who want to be safe from superintelligent AI. Open to all.

• FAR.AI Open Social – London Alignment Workshop 2 March (London, UK).
  Informal evening of networking and conversation hosted by FAR.AI alongside the London Alignment Workshop. Open to anyone interested in AI alignment and related fields, with light refreshments provided.

• Brussels AI Safety Meetup – Europe and the Geopolitics of AGI 23 February (Brussels, Belgium).
  Event hosted by BlueDot Impact exploring the need for a preparedness plan for AGI in Europe. Includes presentations by researchers from the Centre for Future Generations (CFG) and RAND Europe on an AGI preparedness report, followed by Q&A and informal networking.

• Swift Centre AI Policy Challenge Start of March – 20 April (online).
  Open competition translating AI forecasts into policy advice for government decision-makers. Participants analyze forecasts on topics including agentic AI, autonomous weapons, and biosecurity, then develop policy advice in the form of a UK Secretary of State briefing.

• Berkeley ControlConf 2026 18–19 April (Berkeley, USA).
  Conference by Redwood Research and FAR.AI bringing together those interested in or working on AI control to coordinate, share knowledge, and hold discussions. Includes presentations, demos, panels, opportunities for 1-1s, and breakout sessions.

New training programs

• AFFINE Superintelligence Alignment Seminar 28 April – 28 May (Prague, Czechia).
  Month-long program providing promising newcomers to AI alignment a deep understanding of some large pieces of the problem, making them better equipped to work on mitigating AI existential risk. Top candidates may be offered a full-year fellowship.

New events

• PauseCon DC 2026 11–14 April (Washington D.C., USA).
  4-day conference with a mix of workshops, grassroots lobbying, and a chance to meet PauseAI members from across the country. Includes training workshops on how to make political change, and participants will meet with their congressional offices.

• London Alignment Workshop 2–3 March (London, UK).
  Part of a series run by FAR.AI bringing together top ML researchers and practitioners from industry, academia, and government to discuss critical topics related to AI alignment – enabling participants to better understand AGI risks and strategies for solving them.

• Seoul Alignment Workshop 6 July (Seoul, South Korea).
  As above.

• EAGxDC 2026 2–3 May (Washington D.C., USA).
  Effective altruism conference designed to engage the policy, research, and public-interest communities of the Washington D.C., Maryland, and Virginia region. An opportunity to connect with practitioners, explore career paths, and engage with organizations in AI safety.

• EA Summit: Jakarta 19 April (Jakarta, Indonesia).
  Indonesia’s first effective altruism conference, bringing together young professionals, researchers, students, and practitioners from high-impact sectors to explore how they can do the most good with their careers and resources – including in AI safety.

• AI Breakfast: AI Safety as a Startup 17 February (London, UK).
  Morning panel discussion with investors and founders in AI safety, hosted by London AI Hub and Merantix Capital. Aiming to be an open discussion around the challenges facing AI safety founders.

New training programs

None this week.

New events

• MATS Winter Research Talks 6 February (London, UK).
  Past and present MATS fellows in London will be giving talks about their research, followed by Q&A and conversation. Topics will include exploration hacking, misalignment faking, measuring LLM reasoning inconsistency, and model organisms of training hacking.

• EA Summit: Helsinki 28 February (Helsinki, Finland).
  1-day conference for people interested in doing good effectively (including in AI safety) to connect with like-minded individuals and organizations, as well as explore a range of cause areas and the work being done in them.

• AI Safety Connect: 2026 AI India Impact Summit 18–20 February (Delhi, India).
  3-day series of events as part of the India AI Impact Summit. The first day combines technical research and policy coordination, second brings together industry to discuss AI safety norms, and third is a ministerial-level panel on international AI safety coordination priorities.

• PauseAI x Pull the Plug: AI Safety March 28 February (London, UK).
  PauseAI is joining several other groups for the first-ever coalition march of AI activist groups, demanding that the CEOs of every AI company publicly state their support for a pause in principle of AI development.

• Social Choice for AI Ethics and Safety 2026 Europe (SC4AI’26e) 26–27 February (Paris, France).
  Bringing together researchers in AI and formal social choice theory (e.g. voting, preference aggregation, mechanism design etc.) to explore ways in which social choice can help make AI systems more ethical and safe.

New training programs

• Introduction to Cooperative AI (Spring 2026) 9 March – 1 May (online).
  8-week course aiming to deepen participants’ understanding of the field of cooperative AI and prepare them to start or join an ongoing project. Open to diverse backgrounds and career stages – little prior knowledge is expected.

• ILINA Junior Research Fellowship: 2026 early April – end of November (online).
  Designed to immerse participants in the field of AI safety, help them develop high-impact projects, and build career capital. Involves weekly seminar sessions followed by a dedicated research phase with mentorship, split into technical and governance tracks.

• AI Governance Accelerator 14 February – 9 May (Manila, Philippines).
  10-week program run by AI Safety Tiliman for people based in the Philippines and wider Southeast Asia to build foundational knowledge in AI governance. Participants learn how AI systems work, what risks they pose, and how to design governance frameworks.

We’re redesigning the directory and are looking for users to interview. If you’re interested in AI safety and are currently trying to understand what events or programs exist, we’d love to ask you a bunch of (easy) questions to help inform the redesign. Please book a 30 minute call with our designer if you’re keen to help.

New events

• Launch: Safe AI Germany + Virtual Networking 4 March (online).
  BlueDot Impact and SAIGE (Safe AI Germany) are hosting this inaugural Germany virtual AI safety meetup. Open to everyone, from people working in AI safety to those simply curious about the future of AI and its implications.

• AI Safety and Governance Virtual Event | India 4 February (online).
  Meetup aimed at gathering professionals, researchers, and community organizers interested in AI safety initiatives in India before the AI Impact Summit. There will be lightning talks and networking opportunities.

New training programs

A lot this week!

• FIG Fellowship with Yoshua Bengio 7 April – 3 July (online).
  Special round of the Future Impact Group (FIG) Fellowship built around research projects co‑led by Yoshua Bengio. Fellows will work directly with him and senior policy leadership at Mila (Quebec AI Institute) on research to influence the trajectory of frontier AI.

• LASST Summer Fellowship 2026 1 June – 7 August (online).
  10-week summer fellowship with Legal Advocates for Safe Science and Technology (LASST). Fellows work on applied research in AI governance (and biotechnology), participate in strategy development, attend events, and produce a white paper or legal research memo.

• HIP Impact Accelerator Program: Spring 2026 30 March – 17 May (online).
  6-week course run by High Impact Professionals (HIP) helping experienced professionals transition into high-impact careers – like AI safety. Participants follow a guided workbook and attend weekly group sessions, and receive continued support after the program.

• SAIGE Incubator Program: Spring 2026 6 April – 19 July (Germany).
  Inaugural program from the newly-formed Safe AI Germany (SAIGE).

  Over 15 weeks participants will build foundational knowledge, gain hands-on project experience, and connect with mentors and peers across the country.

• ML4Good Spring ‘26: Technical – Western Europe 11–19 April (Lyon, France).
  8-day intensive bootcamp on technical AI safety aimed at people pursuing a career in the field. Curriculum covers technical foundations, AI safety strategy, and technical governance – ending with a capstone project. All backgrounds are welcome.

• ML4Good Spring ‘26: Technical – South Africa 17–25 April (South Africa).
  As above.

• ML4Good Spring ‘26: Technical – Central Europe 6–14 May (Leipzig, Germany).
  As above.

• ML4Good Spring ‘26: Technical – Canada 1–9 June (Montréal, Canada).
  As above.

• ML4Good Spring ‘26: Governance and Strategy – South Africa 17–25 April (South Africa).
  8-day intensive bootcamp on AI governance and strategy, aimed at people pursuing a career in the field. Curriculum covers AI governance, strategy, communication, and career planning – followed by a capstone project. All backgrounds are welcome.

• ML4Good Spring ‘26: Governance and Strategy – Western Europe 4–12 May (Lyon, France).
  As above.

• BlueDot Impact: Technical AI Safety (Feb ‘26) 16 February – 29 March (online).
  6-week course aimed at helping participants understand current AI safety techniques and identify where they can contribute. Each week consists of reading, writing, and a meeting to discuss the material with peers. 6-day intensive version also available.

• BlueDot Impact: Technical AI Safety (Mar ‘26) 16 March – 26 April (online).
  As above.

• BlueDot Impact: Technical AI Safety (Apr ‘26) 6 April – 17 May (online).
  As above.

• BlueDot Impact: AGI Strategy (Feb ‘26) 16 February – 22 March (online).
  5-week course exploring the incentives driving AI companies, what’s at stake, and various strategies for ensuring AI benefits humanity. Each week consists of reading, writing, and a meeting to discuss the material with peers. 5-day intensive version also available.

• BlueDot Impact: AGI Strategy (Mar ‘26) 16 March – 19 April (online).
  As above.

• BlueDot Impact: AGI Strategy (Apr ‘26) 6 April – 10 May (online).
  As above.

• BlueDot Impact: AI Governance (Mar ‘26) 9 March – 12 April (online).
  5-week course aimed at helping participants learn about the policy landscape, regulatory tools, and institutional reforms needed to navigate the transition to transformative AI. Each week consists of reading, writing, and a meeting to discuss the material with peers.

• BlueDot Impact: AI Governance (intensive version) 23–27 February (online).
  5-day course aimed at helping participants learn about the policy landscape, regulatory tools, and institutional reforms needed to navigate the transition to transformative AI. Each day consists of reading, writing, and a meeting to discuss the material with peers.