Existential Risk Taxonomy
Seventeen ER domains in one integrated research model; three proposed categories marked with an asterisk (*).
Explore taxonomy →Risk Assessment & Analysis
Scenario modeling, cascading effects, uncertainty and escalation pathways.
Explore assessment →Research Sources & Attribution
Scholarly bibliography, provenance verification and taxonomy mapping.
Explore sources →Governance, Safety & Societal Responses
Examination of governance frameworks, proposed safety approaches, institutional responses, and international cooperation.
Explore governance →Research caution. Taxonomy entries identify candidate risk mechanisms, not established existential threats. Source associations are preliminary, not endorsements by cited authors. The v0.2 research baseline remains authorized internally; v0.3 is the existing working taxonomy. The v0.4 integrated model includes 17 domains and 112 candidate mechanisms; three proposed categories are marked with an asterisk (*) and gold highlighting. ER-10 already addresses environmental risks and ER-06 already addresses economic and labor risks; these overlaps require explicit resolution. No public publication has been authorized.
17ER domains (3 new candidates*)
112Candidate mechanisms (92 existing + 20 proposed*)
v0.4Integrated editorial prototype
Comprehensive ER Taxonomy
All 17 domains are shown in one integrated model. An asterisk (*) and gold highlighting identify the three proposed categories; their identifiers remain provisional.
Featured Research Scholars
Selected contributors to AI safety and existential-risk scholarship; inclusion does not imply endorsement of IAIGCRM.
Nick Bostrom
Superintelligence (2014) · Long-term AI risk
Stuart Russell
Human Compatible (2019) · AI alignment and control
Roman V. Yampolskiy
AI: Unexplainable, Unpredictable, Uncontrollable (2024) · AI control
Yoshua Bengio
Managing Extreme AI Risks Amid Rapid Progress (2024, coauthor) · AI safety
Geoffrey Hinton
Managing Extreme AI Risks Amid Rapid Progress (2024, coauthor) · Advanced AI risk
Paul Scharre
Army of None (2018) · Autonomous weapons
Research Sources & Attribution
Expanded curated research register. Browse by author, domain or recent publications. All entries require final editorial verification.
02 / Risk Assessment & Analysis — Methodology
Specify threat scenarios, initiating conditions, causal and escalation pathways, systemic impacts, assumptions, uncertainty, confidence and residual risk. Distinguish evidence-based analysis from speculative mechanisms.
04 / Governance, Safety & Societal Responses — Methodology
Examine proposed preventive, detective and responsive safeguards; compare governance arrangements, accountability, oversight and international cooperation. Distinguish demonstrated effectiveness from proposals, unresolved questions and residual uncertainty.
Methodology & Publication Control
Research records distinguish empirical evidence, historical analogy, modeled analysis, theoretical argument and speculation. Consequence tiers (systemic, global catastrophic and existential) are not probability estimates. Capability maturity and uncertainty must be documented independently.
Identifier policy: ER-13 and ER-14 appear immediately after ER-03 in thematic display order. Permanent identifiers do not change. Cross-domain overlaps are retained pending editorial resolution.
Version control: Website v2.3.1 layout retained; integrated taxonomy v0.4 editorial prototype (17 domains, 112 candidate mechanisms, including 3 asterisk-marked new categories / 20 proposed subrisks); source register v0.2 candidate. Publication requires separate authorization.
Research attribution: IAIGCRM research initiative; research leadership Stephen Buerle, PhD. AI-assisted conceptual analysis and documentation: OpenAI ChatGPT (GPT-6), October 25, 2026. Editorial responsibility remains with IAIGCRM.
Overlap review: Candidate B intersects ER-10 Environmental & Planetary Systems; Candidate C intersects ER-06 Economic & Financial Systems; Candidate A intersects ER-11, ER-07 and ER-08. No established IDs have been reassigned. Candidate identifiers are provisional.