Sources & transparency

Annual cog-sci × ML conference. ACO attendees follow for predictive-coding, metacognition, signal-detection-theory measurement work applicable to LLMs.

Cohort program: Cambridge-based fellowship on existential AI risk. Annual application cycle. URL may shift; verify on each scrape.

LessWrong firehose RSS. The legacy /events.rss URL stopped serving XML (returns Next.js HTML shell as of 2026-05). /feed.xml is reachable and stable, matching the alignment-forum pattern. Filter aggressively for events-relevant items.

Annual technical AI safety conference at Oxford. Third iteration. Organized by Oxford Martin AI Governance Initiative and Noeon Research.

Cohort program: multi-week structured AI safety research camp. Multiple iterations per year. Application announcements via lesswrong + their site.

Annual AI governance workshop at AAAI. Third iteration. Focus on alignment, morality, law, and design. Organized by researchers from IBM, ETH Zurich, and other institutions.

Workshop on AI persuasion and manipulation capabilities. Aligned with governance/safety concerns.

12-week AI safety research fellowship funded by Open Philanthropy (Coefficient Giving). Two-stage selection process, 30 fellows selected. Cross-referenced on multiple platforms. Promoted to active after verification.

Alignment Forum firehose RSS. Site /events HTML and tag feeds are persistently 429; only /feed.xml is reachable. Filter inside the agent for events-relevant items.

Anthropic's alignment research blog. High-trust source for safety research and fellowship announcements.

Look for hackathons, fellowships, and ControlConf-style events.

Organizes AI safety research sprints and hackathons. 55+ sprints with 6,000+ participants across 200+ global locations.

AI safety org focused on evaluations and deceptive alignment. Watches for cohort calls and workshop announcements.

Apollo specifically for eval-design and scheming-detection workshops. Distinct from the apolloresearch.ai/blog feed.

Alignment Research Center founded by Paul Christiano in 2021. Berkeley-based nonprofit focused on theoretical AI alignment and mechanistic interpretability. Runs White-Box Estimation Challenge. ARC Evals spun out as METR in 2023.

Cohort program; CFPs run several times a year.

ASSC 29 conference found for 2026 (June 30-July 3 in Santiago, Chile). Promoted to active. Conference relevant for consciousness studies/measurement science applicable to ACO interpretability work.

Promoted to active after confirming Australian AI Safety Forum 2026 event (July 7-8 in Sydney). Interdisciplinary forum grounded in the International AI Safety Report, focusing on measurement, evaluation, and governance of AI systems. Well-established annual event with clear AI safety focus.

UC Berkeley student community for AI safety. Holds Decal course and research program.

UC Berkeley center focused on responsible and secure AI systems. Hosts Agentic AI Summit and other safety-focused events.

Runs AI safety fellowships and AI governance courses. Cohort-based.

BlueDot Impact's Luma calendar featuring AI safety events, reading clubs, meetups, and hackathons. JS-heavy, use web_fetch_rendered.

Boston community group for AI safety discussions. Regular meetups.

Cohort program: 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, MA.

Cambridge, MA-based 501(c)(3) non-profit founded in 2022. Runs research fellowships in AI alignment and governance.

CBAI's programs index, listing current cohorts (summer research fellowship, AIxBiosecurity fellowship, etc.). Auto-updates each year as new programs land; year-specific URLs go stale.

Workshops, hackathons, ML Safety Newsletter cross-references.

Cohort program: annual internship for grad students. Stuart Russell's lab at Berkeley. High-leverage application window.

Constellation's Astra Fellowship program. Fellows work on AI safety projects with expert mentorship. Strong placement rates at safety orgs.

Annual conference on AI control organized by Redwood Research and FAR.AI. Focuses on reducing risks from misalignment through safeguards that work even when AI models are trying to undermine them.

Mixed AI-safety and EA-broad events. Filter aggressively.

EA Global conference series organized by Centre for Effective Altruism. Heavy AI safety attendance and programming. In scope for social/community event reasons.

EleutherAI mech-interp working group. Active reading-group + paper-discussion cadence.

Coalition focused on AI evaluation practices. Runs workshops at major conferences.

Foundational AI Research organization. Runs workshops on AI safety verification, secure compute, and other technical safety topics. FAR Labs provides coworking space and events for Bay Area AI safety community.

Bay Area coworking + event venue for AI safety / interpretability researchers. Regular talks and reading groups.

Federation of American Scientists AI safety policy fellowship program. Found via discovery search.

Cohort program: fellowship for students and early-career researchers in AI governance, technical AI safety, and digital sentience. Associates dedicate 8+ hours/week. FIG is an acronym with potential name collision; requires careful verification.

Foresight Institute hosts Vision Weekend events with frontier science and technology tracks including AI safety. 40-year-old organization focused on transformative technology.

Sociotechnical risk and threat-actor analysis for advanced AI. Source for counter-autonomous / facilitation-forensics events.

Mechanistic interpretability research org. Watches for SAE / probe / circuit tracing announcements.

Filter to safety/alignment specifically.

Centre for the Governance of AI. Runs summer fellowships in DC and London focused on AI governance and policy. High-trust research organization.

Annual; cross-reference workshop list with safety keywords.

Annual theoretical AI alignment conference. Unconference format. Organized by Iliad, an umbrella organization for applied mathematics research in alignment.

First global academic programme focused on AI evaluations. Funded by Coefficient Giving. Faculty from Cambridge, Stanford, Princeton, EU AI Office, UK AI Safety Institute, FAR AI, Apollo Research.

Lu.ma is the de-facto calendar for safety community informal events (mixers, salons, hackathons). The lu.ma → luma.com migration happened 2026-05; URL pinned to canonical destination so we don't redirect-hop on every fetch. JS-heavy: fetch with web_fetch_rendered.

Lu.ma alignment-tagged events. Pinned to luma.com after the 2026-05 lu.ma → luma.com migration. Use web_fetch_rendered.

MIT student group conducting AI alignment research. Runs reading groups and ML bootcamps. Supported by CBAI. Membership in hundreds.

Cohort program: matches exceptional students and early-career researchers with experienced researchers from AI labs, think tanks, and academia. Operated by Cambridge AI Safety Hub. Multiple cohorts per year.

Fellowship cohort and CFPs.

Annual mechanistic interpretability workshop at ICML. High-quality venue for mech interp research. Organized by leading researchers in the field.

Evaluations org. Watches for hires + workshop announcements.

Often the first to mention new workshops and CFPs.

Annual; cross-reference workshop list with safety keywords.

Regional mechanistic interpretability workshop series for New England researchers. Found via discovery search.

OpenAI's safety fellowship program, partnering with Constellation. High-trust source for AI safety research fellowships.

Cohort program: fellowship pairing safety researchers with formal-cognitive-science mentors. Rebranded from PIBBSS to Principles of Intelligence. Domain changed from pibbss.ai to princint.ai (301 redirect detected 2026-05-02). Runs 3-month fellowship, affiliate program, and Iliad Research Residency.

AI safety research fellowship based in London. Quarterly cohorts.

Annual workshop at ICML on pluralistic AI alignment. Focus on integrating diverse perspectives into AI alignment frameworks.

First Workshop on Pluralistic Value Alignment of LLMs at AACL 2026 in Zhuhai, China. Cross-referenced via ACL web portal.

Annual workshop on representational alignment at ICLR. Focus on alignment among artificial and biological information processing systems.

Control / interpretability focus. Watches for ControlConf, REMIX, etc.

Cross-disciplinary conference on decision making, learning, cognition. Increasingly hosts LLM-cog-sci crossover work.

Luma calendar for Singapore AI Safety Hub community events. JS-heavy, use web_fetch_rendered. Currently no upcoming events but active calendar infrastructure.

Cohort program: part-time remote research program pairing aspiring researchers with experienced mentors from Google DeepMind, RAND, Apollo Research, MATS, UK AISI, etc. for three-month projects. Multiple cohorts per year (Fall, Spring). Well-established programme with broad mentor base.

Confirmed legitimate Stanford AIMS Lab workshop with high-quality venue (COLM workshop). Promotes to active.

10-week hybrid fellowship in Sydney, Australia. Cross-referenced on EA Forum and Alignment Forum. Organized by Chris Leong and Jack Payne. Promoted to active after verification.

UK government AISI. Hosts evaluation workshops and publishes calls for collaboration.

United Nations Institute for Disarmament Research. International AI policy and disarmament-related conferences in scope per the safety community's broader governance interests.