BlueDot Impact launch event in San Francisco bringing together the AI safety community. Part of BlueDot's expansion to build the workforce needed to safely navigate AGI. Community gathering for networking and connection among local AI safety practitioners, researchers, and fellows.
Past events
Sorted by date, newest first. Useful as a memory of what's happened in the community.
Part 2 of the Breaking Barriers to AI Safety series, featuring a career fair connecting AI safety job seekers with organizations, plus a builders summit for technical collaboration. Organized by Jen Ba and Robin Goins (Mox) through BlueDot Impact.
Apart Research hackathon focused on AI safety research. Part of the 55+ sprints series with 6,000+ participants across 200+ global locations.
Weekly AI safety evaluations paper reading club organized by BlueDot Impact. Virtual session on Zoom. Part of BlueDot's regular community programming building the workforce needed to safely navigate AGI.
Part 1 of a two-part AI safety event series organized by BlueDot Impact. This hackathon focuses on building AI safety tools and breaking barriers to entry in the field. Followed by Part 2 (Career Fair and Builders Summit) on July 24. Organized by Joshua Landes and Jen Ba.
Technical workshop bringing together top talent to address bottlenecks in secure AI advancement. Organized by Foresight Institute, a 40-year-old organization focused on transformative technology. Registration available.
A community event organized by Singapore AI Safety Hub (SASH) focused on getting into AI safety. Provides networking and information for those interested in entering the AI safety field.
Second Workshop on Agents in the Wild focusing on safety and security of AI agents deployed in real-world environments. Addresses challenges in ensuring safe and secure operation of autonomous agents. Part of ICML 2026 workshop track.
The workshop addresses how to integrate diverse perspectives, values, and expertise into pluralistic AI alignment frameworks. Examines multi-objective alignment approaches, preference elicitation methods, and human-AI interaction workflows that reflect pluralistic values across diverse communities.
Workshop addressing Pluralistic AI: Aligning with the Diversity of Human Values. Examines how to integrate diverse perspectives into AI alignment frameworks, exploring multi-objective approaches and consensus-building practices for navigating value conflicts in pluralistic societies.
Second Workshop on Technical AI Governance Research at ICML 2026, focusing on technical approaches to AI governance, policy, and regulation. Part of the main conference workshop track.
Workshop at ICML 2026 focused on identifying, diagnosing, and fixing failure modes in agentic AI systems. Covers reproducible triggers for failures, diagnostic tracing methods, and verified repair approaches. Highly relevant to AI safety and robustness.
Workshop bringing together diverse perspectives from the community to discuss recent advances in mechanistic interpretability, build common understanding and chart future directions. Addresses developing principled methods to analyze and understand model internals (weights and activations) to gain insight into behavior and underlying computation. Received 2.6x submissions from previous year.
Annual mechanistic interpretability workshop at ICML building community dialogue around understanding neural network internals through principled analysis methods. Features 23 spotlight presentations alongside poster presentations. Continues series from previous workshops at ICML 2024 and NeurIPS 2025.
Annual mechanistic interpretability workshop at ICML. High-quality venue for mech interp research organized by leading researchers in the field.
An interdisciplinary forum grounded in the science of AI safety, bringing together participants from research, government, industry, and civil society. The 2026 forum features 79 speakers across 76 sessions with 329 participants, focusing on measurement, evaluation, and governance of AI systems.
3-week ML upskilling bootcamp for AI safety focusing on interpretability and RL, based on ARENA curriculum. Run by Cambridge Boston Alignment Initiative. Provides housing, meals, 24/7 office access, dedicated teaching assistants, and travel support to participants.
International Conference on Machine Learning. July 6 for Expo/Tutorial Day, July 7-9 for Main Conference, July 10-11 for Workshops. Annual ML conference with safety-related workshops in scope.
Part of the ongoing Alignment Workshop series by FAR.AI, bringing together global leaders to explore strategies for mitigating risks from artificial general intelligence. FAR AI is a foundational research organization focused on AI safety verification, secure compute, and technical safety topics.
AI safety conference in Christchurch, New Zealand. Organized by BlueDot Impact community. In-person event bringing together the New Zealand AI safety community.
Workshop examining tensions between model developers and evaluation researchers, covering evaluation methodology and measurement theory, evaluation infrastructure and costs, and sociotechnical impact assessments. Co-hosted social with GEM workshop on July 3 around 8pm, main workshop sessions on July 4 afternoon.
29th Annual Meeting of the Association for the Scientific Study of Consciousness in Santiago de Chile. Academic society promoting rigorous research on understanding the nature, function, and underlying mechanisms of consciousness. Relevant for ACO practitioners working on consciousness studies and measurement science applicable to interpretability work. Includes members from cognitive science, medicine, neuroscience, philosophy, and related disciplines.
AI policy hackathon focused on governance challenges of deployed AI systems. Organized by BlueDot Impact community in Toronto, bringing together participants to work on practical policy frameworks for AI governance in real-world deployment contexts.
Regional AI safety hackathon for Latin America, Africa, and Asia. Participants build AI safety tools, evaluations, and policy research. Regional competition structure with pipeline to fellowship and placement opportunities. Funded by Schmidt Sciences.
United Nations Institute for Disarmament Research global conference on AI governance, security, and ethics. Part of UNIDIR's Centre of Excellence on AI, Peace and Security programming. International AI policy and disarmament-related conference in scope per the safety community's broader governance interests.
Two-week intensive summer school at Columbia University covering machine learning topics including mechanistic interpretability, alignment/safety, RAG & agents, and LLM systems. Approximately 200 PhD students participate alongside faculty and industry speakers. In-scope due to dedicated alignment and mechanistic interpretability tracks.
Bringing together multiple faith perspectives on AI governance and security implications, exploring ethical dimensions of advanced AI through inter-religious dialogue.
AI Risk Content Hackathon organized by BlueDot Impact in London. Focus on creating AI safety and risk communication content. Part of BlueDot's broader mission to build the workforce needed to safely navigate AGI.
Foresight Institute flagship event gathering leading scientists, entrepreneurs, funders, and policymakers to explore the frontiers of science and technology. Multiple tracks including AI safety topics.
Flagship conference where leading scientists, entrepreneurs, funders, and policymakers convene to explore frontier technology and plan for beneficial futures. Includes AI safety track as part of broader focus on transformative technology. 40-year-old organization focused on beneficial technology development.
Foresight Institute Vision Weekend with frontier science and technology tracks including AI safety. 40-year-old organization focused on transformative technology. Three-day event in London featuring AI safety programming alongside other frontier tech tracks.
Five-day intensive programme for AI safety founders going from idea to funded. Successful pitches receive £50k in equity-free seed funding. Part of BlueDot Impact's incubator and rapid-funding initiatives supporting concrete AI safety work.
EA Global conference series organized by Centre for Effective Altruism. Speakers present research on effective altruism including heavy AI safety programming. Features talks, workshops, and networking opportunities. In scope for social and community event reasons with significant AI safety attendance.
In-person bootcamp covering AI safety fundamentals, mechanistic interpretability, and reinforcement learning. Program covers travel, visas, accommodation, and meals. Duration: 4-5 weeks typical for ARENA bootcamps. Online curriculum available for independent study.
8-week virtual reading group run by MIT AI Alignment (MAIA), meeting 2 hours per week. Explores why AI safety matters and current mitigation approaches including AI trajectory, misalignment risks, technical safety solutions, policy, and career paths. No prior AI background required. Led by small groups facilitated by MAIA team members.
Hackathon focused on secure program synthesis by Apart Research. Hybrid format with online participation and in-person hubs.
5-week project-based course 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. Covers alignment, mechanistic interpretability, evaluations, red-teaming, AI control, and scalable oversight.
3-week ML upskilling bootcamp for AI safety focusing on interpretability and RL, based on ARENA curriculum. Run by Cambridge Boston Alignment Initiative. Provides housing, meals, 24/7 office access, dedicated teaching assistants, and travel support to participants.
FAR.AI and Center for AI Safety workshop on infrastructure for secure and verifiable AI, bringing together researchers, builders, and funders across ML, hardware security, systems, cryptography, and computer security to identify the most promising technical approaches and spark concrete collaborations.
Free, one-day technical AI safety conference organized by Oxford Martin AI Governance Initiative and Noeon Research. Third iteration. Welcomes researchers and professionals from all backgrounds interested in discussing AI safety, regardless of prior experience. Sponsored by MATS and Apart Research.
A free, accessible workshop hosted by AI Safety Awareness Group Oakland exploring AI's trajectory and societal impact. No technical background required. Features live demonstrations of current AI systems, interactive forecasting activities, and discussions about AI's implications for work, relationships, and society over the next 1-5 years.
Two-day conference on global cooperation for cybersecurity resilience and stability. Organized by United Nations Institute for Disarmament Research. Addresses international frameworks for cyber governance and security cooperation.
13-week part-time remote fellowship designed to develop Black researchers, practitioners, and leaders in AI Safety, AI Security, and AI Governance. Program runs in two phases: Weeks 1-5 cover foundations and training curriculum, while Weeks 6-13 focus on research and project development. Organized by Black in AI Safety & Ethics (BASE) to empower Black researchers in the AI safety community.
A community-organized regional Effective Altruism conference targeting the policy, research, and public-interest communities across the Washington DC, Maryland, and Virginia area. Designed to help professionals connect with practitioners, explore high-impact career paths, and engage with organizations focused on global challenges. Key topics include global health, AI governance, public policy, and biosecurity.
AI safety mixer for professionals exploring a move into AI safety, hosted by the London Initiative for Safe AI (LISA). A low-pressure evening to explore the field, understand the part you could play, and meet people who are already working in AI safety. Designed for professionals interested in transitioning into the AI safety field.
Third edition workshop bringing together researchers from machine learning, neuroscience, and cognitive science to explore representational alignment among artificial and biological information processing systems. This year focuses on what we can do with alignment, emphasizing practical affordances and how alignment transforms static representations into controllable computational primitives.
Three-day hybrid hackathon focusing on AI and biosafety. Organized by Apart Research with in-person hubs in London, Berlin, and San Francisco.
Conference on AI control, focusing on reducing risks from AI misalignment through interventions that are robust even when AI models involved are attempting to undermine those safeguards. Features talks, fireside chats, and breakout discussions with experienced researchers. Free to attend, with a pre-conference workshop on April 17. Organized by Redwood Research & FAR.AI.
Second annual Technical Innovations in AI Policy Conference organized by FAR.AI in collaboration with leading think tanks. Focuses on technical innovations that enable AI policy implementation.
A three-day hackathon focused on developing safety measures for potentially misaligned AI systems. Participants work on control protocols and evaluation tools to keep autonomous AI agents contained, addressing oversight challenges as AI systems become more autonomous. Co-organized by Apart Research and Redwood Research (founder of the AI control field), with 700+ participants submitting 126 projects across main competition and specialty tracks.
FAR.AI hosted event bringing together more than 200 researchers, policymakers, and industry experts to discuss AI alignment topics. Featured presentations on honeypot-based methods for detecting scheming, and training against interpretability-based deception detectors. Victoria Krakovna (Google DeepMind) and Stefan Heimersheim (Google DeepMind) presented.
The inaugural AIMII workshop brings together researchers across disciplines to examine how generative AI models influence information creation and access, clarifying core concepts, evaluating evidence on AI's persuasive and manipulative capabilities, and exploring implications for society and democracy.
Half-day workshop at IASEAI'26 bringing together researchers from computer science, cognitive science, philosophy, political science, and policy to examine AI's manipulative capabilities and information integrity concerns. Features three panel discussions with leading researchers, plus a poster session showcasing 24 accepted posters and 5 hackathon-winning projects addressing definitions and taxonomies of persuasion, manipulation, and deception.
A part-time, remote research fellowship enabling aspiring AI safety and policy researchers to collaborate with professionals on impactful projects addressing risks from artificial intelligence. Participants commit 5-40 hours weekly for approximately 3 months, culminating in a Demo Day presentation. Mentors include professionals from Google DeepMind, RAND, Apollo Research, MATS, UK AISI.
SPAR (Supervised Program for Alignment Research) Spring 2026 mentee track. Part-time remote research program pairing aspiring researchers with experienced mentors from Google DeepMind, RAND, Apollo Research, MATS, UK AISI for three-month projects. Mentees commit 5-40 hours per week. Research period: February 16 - May 16. Mentee decisions were sent out February 2-6. Applications for Spring 2026 have closed.
Part-time remote research fellowship pairing aspiring researchers with 130+ experienced mentors from Google DeepMind, RAND, Apollo Research, MATS, UK AISI, and other top organizations for three-month AI alignment projects. Participants commit 5-40 hours weekly. Covers project expenses including compute and API/LLM access. Culminates in virtual Demo Day with prizes totaling $7,000. Optional continuation beyond May 16. Mentor application track: experienced researchers from Google DeepMind, RAND, Apollo Research, MATS, UK AISI etc. apply to mentor a project. Mentor application deadline 2025-12-05 (passed).
A three-day conference bringing together the effective altruism community to share new thinking and research, coordinate on global projects, and network. Features keynote speakers, talks, workshops, and social activities focused on addressing pressing global problems including AI safety. Friday evening opening reception, full-day Saturday and Sunday programming.
A fully-funded academic programme on AI evaluation combining technical training with policy and governance perspectives. 40 scholars globally receive 90 hours online, 20 hours hands-on courses, and a 40-hour in-person capstone week in Valencia, earning a 15 ECTS Expert Diploma via ValgrAI.
A three-day hackathon bringing together builders to create verification systems, compliance infrastructure, and coordination tools for international AI governance. Five focus tracks: Hardware Verification & Attestation, Compliance Infrastructure & Privacy-Preserving Proofs, Risk Thresholds & Compute Verification, International Verification & Coordination, and Research Governance & Dual-Use Detection. Organized by Apart Research in partnership with MIRI Technical Governance Team and Lucid Computing.
The 3rd International AI Governance Workshop focuses on Alignment, Morality, Law, and Design. This full-day hybrid workshop brings together researchers, industry, and policy communities to bridge technical alignment methods, policy frameworks, and ethical implementation of AI governance.
Twelve-week intensive AI safety research fellowship with 25+ hours per week commitment. Two-stage selection: 60 participants accepted for trial week (Jan 19-23) on foundational coursework in RLHF, interpretability, SAEs, and adversarial robustness, then 30 selected as Research Fellows based on performance. Fully funded by Open Philanthropy, covers tuition, computing infrastructure access, mentorship, and limited conference travel support.
Three-month online AI safety research program where participants form teams to work on pre-selected projects. Opening weekend January 10-11, projects run through April 19, final presentations April 24-27. Features 27 projects across six themes: Stop/Pause AI (5), Policy/Governance (5), Evaluate Risks from AI (5), Mech-Interp (2), Agent Foundations (4), Alternative LLM Safety (5), and Safe by Design AIs (1). Participants work 10 hours per week with weekly team meetings. Seven-year track record with alumni founding 10 organizations and securing 43 jobs in AI Safety. Application deadline November 23, 2025.
3-month long online program where participants form teams to work on pre-selected AI Safety projects. Opening weekend January 10-11 through April 19 with final presentations April 24-27. Three tracks available covering alignment research, governance, policy, and stop/pause advocacy. Team member applications deadline was November 23rd.
10-week hybrid fellowship in Sydney, Australia. 7 weeks in-person component, 2 days per week in-person attendance. Targets strongly motivated, highly agentic, immensely talented folks across diverse backgrounds from technical researchers to governance specialists to entrepreneurs. Focuses on helping participants develop strategic awareness and identify their optimal contribution path to AI safety. Weekly customized discussions, expert speakers, mentorship, co-working space with compute access, social events, potential flight reimbursement for top candidates.
Cambridge Bootcamp for Research in Interpretability and Alignment: 3-week ML upskilling bootcamp for AI safety, focusing on interpretability and RL. Based on ARENA curriculum. Run by Cambridge Boston Alignment Initiative in Cambridge, Massachusetts. In-person intensive programme.
Part-time, remote-first, 12-week fellowship by Future Impact Group for students and early-career researchers. Minimum 8+ hours/week on research projects in AI policy, philosophy for safe AI (technical safety and ethical foundations), or AI sentience. Provides co-working sessions, issue troubleshooting, career guidance, networking, and guest speakers.
Part-time, remote-first, 12-week research fellowship where participants work as research associates on specific projects under experienced supervision. Focus areas: AI governance, technical AI safety, and digital sentience. Time commitment: 8+ hours per week. Includes co-working sessions, issue troubleshooting, career guidance, opening and closing events, networking opportunities, research sprints, and guest speakers.