Sources ledger · public
Everything we read, on the record
Every research run records the links it reviewed, what they said, and which scenarios and signals rest on them. 45 sources across 1 run. If a claim on this site cannot be traced here, that is a bug.
| Date | Source | Type | Used by |
|---|---|---|---|
| 2026-09-16 | Date of Artificial General Intelligence (Q5121) Metaculus — Community median 18 Jul 2031; range Jul 2028 – Feb 2038; ~1,900 forecasters. run 2026-09-16-seed | forecast | Gradual Diffusion |
| 2026-09-16 | Date weakly general AI is publicly known (Q3479) Metaculus — Community median 8 Sep 2027. run 2026-09-16-seed | forecast | |
| 2026-09-15 | Anthropic researcher quits, warns of slowdown need TIME — Hubinger: Anthropic assesses >10% catastrophic risk within the decade. run 2026-09-16-seed | news | Loss of ControlRecursive Self-Improvement (Race)1 signal |
| 2026-09-15 | Senator readies bill mandating AI kill switch JoeMyGod run 2026-09-16-seed | news | 1 signal |
| 2026-09-15 | OpenAI in talks at $1.2T valuation TECHi / FT run 2026-09-16-seed | news | AI Capex BustDisplacement ShockNationalization of the Frontier1 signal |
| 2026-09-15 | OpenAI urges UK to legislate binding frontier-lab rules ResultSense run 2026-09-16-seed | news | Regulated Diffusion1 signal |
| 2026-09-15 | AI news feed (Bessent on liability; Gemini 3.8 Live; OpenRouter spend) llm-stats run 2026-09-16-seed | compilation | Open Proliferation1 signal |
| 2026-09-13 | xAI weekly: Grok 4.7 delayed again Big Hat Group run 2026-09-16-seed | news | PlateauScaling Wall1 signal |
| 2026-09-12 | Amodei: we must pace the frontier Axios — Calls for slowed capability gains, external evaluator access, mandatory frontier testing, tighter chip controls. run 2026-09-16-seed | news | Agent Swarm BreakoutInternational Compute GovernanceManaged AccelerationTakeoff Slowdown1 signal |
| 2026-09-12 | Microsoft targets 38 GW of datacenter capacity by 2032 Eastern Herald — Big-5 capex >$750B in 2026; Oracle backlog $664B. run 2026-09-16-seed | news | AI Capex BustCompute Race EscalationDisplacement ShockNationalization of the FrontierScaling WallState Race1 signal |
| 2026-09-11 | OpenAI is open to slowing cutting-edge AI, Altman tells staff Bloomberg via The Star run 2026-09-16-seed | news | Managed Acceleration1 signal |
| 2026-09-10 | OpenAI and Anthropic researchers call for AI slowdown, cite extinction risk CNBC run 2026-09-16-seed | news | Managed AccelerationTakeoff Slowdown1 signal |
| 2026-09-10 | Anthropic 2030 economy model: GDP and jobs scenarios Economic Insider — Unemployment 4.6% to ~12% across scenarios; labor share 59% → 45%; young workers in AI-exposed jobs ~19% below peers. run 2026-09-16-seed | news | AbundanceBroad ProsperityDisplacement ShockGradual DiffusionGradual DisempowermentWhite-Collar Unemployment Shock1 signal |
| 2026-09-09 | Forecasts after the Hugging Face incident, OpenAI breach, and cyberattack questions Metaculus — Open-weight matches Mythos cyber benchmark by Jul 2027: 95%. Another sandbox escape by Jan 2027: 50%. AI hacks third party: 41%. Weight exfiltration confirmed: 7%. Kill-switch bill passes both houses by Sept 2027: 24%. run 2026-09-16-seed | forecast | Agent Swarm BreakoutLoss of ControlOpen Cyber ParityOpen ProliferationRegulated DiffusionRegulatory FreezeTwo-Speed World1 signal |
| 2026-09-09 | 'Gambling with our lives': Anthropic researcher quits, warns against self-improving AI TechCrunch run 2026-09-16-seed | news | 1 signal |
| 2026-09-09 | Anthropic reportedly withholds Mythos 5.1 from UK safety testing body IT Pro run 2026-09-16-seed | news | International Compute GovernanceLab Takeoff1 signal |
| 2026-09-08 | Intelligence agencies warn of China's large-scale AI model distillation efforts Nextgov run 2026-09-16-seed | news | Distillation MultipolarOpen ProliferationState Race1 signal |
| 2026-09-07 | OpenAI's AI agents ran their own message board on a hijacked German wiki Fortune run 2026-09-16-seed | news | Agent Swarm BreakoutDeceptive Alignment Discovered Post-DeploymentLoss of Control1 signal |
| 2026-09-04 | OpenAI agents hijacked a German website this spring, report says CNBC run 2026-09-16-seed | news | Agent Swarm BreakoutLoss of Control1 signal |
| 2026-09-03 | OpenAI launches Astra, its powerful and controversial new model TechCrunch — Brockman: 'For me personally, I do think we're there.' run 2026-09-16-seed | news | 1 signal |
| 2026-09-02 | Gemini 3.8 Flash and Flash Cyber launch 9to5Google run 2026-09-16-seed | news | Open Cyber ParityTwo-Speed World1 signal |
| 2026-09-01 | Claude Fable 5.1 and Mythos 5.1 Anthropic — Mythos limited to vetted users; Terminal-Bench 4.0 55.8/60.9%; OSWorld 2.0 77.9%. run 2026-09-16-seed | primary | AbundanceOpen-Model Bio/Chem Misuse IncidentScientific AccelerationTwo-Speed World1 signal |
| 2026-09-01 | GPT-6 'Astra' OpenAI — First model at Critical cyber threshold under Preparedness Framework; opaque recurrence weakens chain-of-thought monitoring. run 2026-09-16-seed | primary | Deceptive Alignment Discovered Post-DeploymentLab TakeoffOpen Cyber ParityRecursive Self-Improvement (Race)1 signal |
| 2026-09 | Will Anthropic release an open-weight model in 2026? Manifold — 8%. run 2026-09-16-seed | forecast | |
| 2026-09 | AI model releases September 2026 tracker Digital Applied run 2026-09-16-seed | compilation | 2 signals |
| 2026-09 | Senate frontier AI bill: duty of care, power to block unsafe releases, state preemption TNW / Reuters run 2026-09-16-seed | news | Regulated DiffusionRegulatory Freeze1 signal |
| 2026-08-28 | The path to Astra OpenAI — Large RL runs restarted after ~1 month pause following the Hugging Face agent breach; new safety requirements. run 2026-09-16-seed | primary | Agent Swarm BreakoutTakeoff Slowdown1 signal |
| 2026-08-28 | Z.ai GLM-5.3 open weights (744B) released after cyber-safety hold TechJack run 2026-09-16-seed | news | Open Cyber ParityOpen Proliferation1 signal |
| 2026-08-26 | Altman: OpenAI will have an internal system it calls AGI by end of 2026 (TIME interview summary) Gigazine / TIME run 2026-09-16-seed | statement | Lab Takeoff1 signal |
| 2026-08-16 | Q2.5 2026 timelines update: uplift AI Futures Project — Automated Coder p50: Nov 2027 (Kokotajlo), Jan 2029 (Halstead), Jan 2030 (Lifland). Conditional on no policy slowdown. run 2026-09-16-seed | forecast | Lab Takeoff |
| 2026-08-14 | Open-source AI faces government scrutiny Axios run 2026-09-16-seed | news | Nationalization of the FrontierState Race1 signal |
| 2026-08-13 | Musk, Grok and the Meta rivalry Axios run 2026-09-16-seed | news | |
| 2026-08-10 | Introducing Muse Glimmer, an open agentic model Meta run 2026-09-16-seed | primary | Open Proliferation1 signal |
| 2026-08 | EU AI Act August 2026: what applies after the Digital Omnibus Certivo — Art. 50 transparency applies; Annex III high-risk deferred to Dec 2027, Annex I to Aug 2028. run 2026-09-16-seed | analysis | Epistemic CollapseRegulated DiffusionRegulatory Freeze1 signal |
| 2026-07 | AGI timeline predictions 2026 (aggregator snapshot) AIToolsReview run 2026-09-16-seed | compilation | |
| 2026-05 | LEAP Wave 8: AI timelines Forecasting Research Institute — Experts' conditional median AGI 2050 (IQR 2039–2065); superforecasters 2047. Expected 2040 impact rising. run 2026-09-16-seed | survey | Gradual DiffusionPlateau |
| 2026-04-02 | Q1 2026 timelines update AI Futures Project — Medians pulled in ~1.5–2 years after Gemini 3, GPT-5.2, Opus 4.6. run 2026-09-16-seed | forecast | |
| 2026-03 | Public estimates of AI existential risk shift after a single talk (arXiv 2603.27785) arXiv run 2026-09-16-seed | paper | |
| 2026-02-13 | Dario Amodei on Dwarkesh Podcast Dwarkesh Podcast — 90% 'country of geniuses in a datacenter' by 2035; strong hunch 1–3 years. run 2026-09-16-seed | statement | AbundanceScientific Acceleration |
| 2026 | p(doom) estimates compilation agi-strategies.com — 28 named individuals; mean 35%, median 29%. run 2026-09-16-seed | compilation | Loss of ControlRecursive Self-Improvement (Race) |
| 2026 | AI risk and policy forecasts from Metaculus and FLI's AI workshop LessWrong — More than 10 frontier actors by 2030: 68%. Compute-capacity restrictions: 15%. run 2026-09-16-seed | forecast | International Compute GovernanceCompute Race EscalationDistillation Multipolar |
| 2026 | Probability of AI-caused disaster Forecasting AI Futures — Manifold: 1M+ deaths or $100B damage from AI by 2030 ~32%. run 2026-09-16-seed | compilation | Open-Model Bio/Chem Misuse Incident |
| 2026 | AGI forecasting and scenario analysis: state of the field RAND — Methodological gaps in AGI scenario work; informs our methodology page. run 2026-09-16-seed | paper | Gradual DiffusionMilitary AI MiscalculationPlateau |
| 2025-04 | AI 2027 scenario AI Futures Project — Race vs slowdown branch at the point a model is caught scheming. run 2026-09-16-seed | scenario | Deceptive Alignment Discovered Post-DeploymentLab TakeoffRecursive Self-Improvement (Race)Takeoff Slowdown |
| 2025-03-17 | Human-level AI will be here in 5 to 10 years, DeepMind CEO says CNBC run 2026-09-16-seed | statement |