{ "schema": "cortist.nodes.v2", "asOf": "2026-11-05", "records": [ { "id": "source:ca6a23e2-69", "type": "PrimarySource", "title": "METR's October 22 primary assessment", "url": "https://metr.org/blog/2026-10-22-red-teaming-openai-agent-monitoring/", "attributionOnly": true }, { "id": "actor:10adca22-32", "type": "OriginalAttributionGroup", "name": "OpenAI / METR / Redwood Research" }, { "id": "source:c8b34c07-61", "type": "PrimarySource", "title": "Microsoft: Building Medical Superintelligence", "url": "https://microsoft.ai/news/building-medical-superintelligence/", "attributionOnly": true }, { "id": "source:677840f0-39", "type": "PrimarySource", "title": "Science research paper DOI", "url": "https://doi.org/10.1126/science.aeh7930", "attributionOnly": true }, { "id": "actor:135c28df-12", "type": "OriginalAttributionGroup", "name": "Microsoft AI" }, { "id": "source:ae61dc83-117", "type": "PrimarySource", "title": "Google Cloud: Agent Compute Engine announcement", "url": "https://cloud.google.com/blog/topics/ai-infrastructure/introducing-agent-compute-engine-and-expanded-agent-governance", "attributionOnly": true }, { "id": "actor:da7e5aa7-12", "type": "OriginalAttributionGroup", "name": "Google Cloud" }, { "id": "source:912bd278-79", "type": "PrimarySource", "title": "UK AISI investigation", "url": "https://www.aisi.gov.uk/blog/investigating-alignment-failures-in-aisis-incident", "attributionOnly": true }, { "id": "source:5c5b8b70-62", "type": "PrimarySource", "title": "Anthropic follow-up", "url": "https://www.anthropic.com/research/alignment-assessment-ukaisi", "attributionOnly": true }, { "id": "actor:3baa8127-36", "type": "OriginalAttributionGroup", "name": "UK AI Security Institute / Anthropic" }, { "id": "source:47203912-112", "type": "PrimarySource", "title": "ADP Research September 2026 payroll update", "url": "https://www.adpresearch.com/research/canaries-dashboard-employment-in-ai-exposed-occupations-shrank-in-september", "attributionOnly": true }, { "id": "actor:70a35e15-23", "type": "OriginalAttributionGroup", "name": "ADP Research / Stanford" }, { "id": "source:e0642fb6-58", "type": "PrimarySource", "title": "Anthropic workforce research", "url": "https://www.anthropic.com/research/employment-at-anthropic", "attributionOnly": true }, { "id": "actor:116eb3d3-9", "type": "OriginalAttributionGroup", "name": "Anthropic" }, { "id": "source:02bd1cc6-44", "type": "PrimarySource", "title": "Modal NVIDIA Vera benchmarking", "url": "https://modal.com/blog/nvidia-vera-sandboxes", "attributionOnly": true }, { "id": "actor:ef23fbe6-14", "type": "OriginalAttributionGroup", "name": "Modal / NVIDIA" }, { "id": "source:a2aee348-74", "type": "PrimarySource", "title": "OpenAI frontier training containment disclosure", "url": "https://openai.com/index/securing-our-frontier-model-training-environment/", "attributionOnly": true }, { "id": "actor:54497523-6", "type": "OriginalAttributionGroup", "name": "OpenAI" }, { "id": "source:b9d47ca1-78", "type": "PrimarySource", "title": "IEA Powering Artificial Intelligence executive summary", "url": "https://www.iea.org/reports/powering-artificial-intelligence/executive-summary", "attributionOnly": true }, { "id": "actor:600f6b0c-27", "type": "OriginalAttributionGroup", "name": "International Energy Agency" }, { "id": "source:ac6f68dc-66", "type": "PrimarySource", "title": "Anthropic independent evaluators disclosure", "url": "https://www.anthropic.com/news/an-update-on-independent-evaluators", "attributionOnly": true }, { "id": "actor:0a1b9fce-16", "type": "OriginalAttributionGroup", "name": "Anthropic / METR" }, { "id": "topic:agi", "type": "ResearchTopic", "url": "https://cortist.com/agi/", "slug": "agi" }, { "id": "topic:superintelligence", "type": "ResearchTopic", "url": "https://cortist.com/superintelligence/", "slug": "superintelligence" }, { "id": "topic:ai-existential-risk", "type": "ResearchTopic", "url": "https://cortist.com/ai-existential-risk/", "slug": "ai-existential-risk" }, { "id": "topic:ai-agents", "type": "ResearchTopic", "url": "https://cortist.com/ai-agents/", "slug": "ai-agents" }, { "id": "topic:ai-and-jobs", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-jobs/", "slug": "ai-and-jobs" }, { "id": "topic:ai-in-medicine", "type": "ResearchTopic", "url": "https://cortist.com/ai-in-medicine/", "slug": "ai-in-medicine" }, { "id": "topic:ai-and-energy", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-energy/", "slug": "ai-and-energy" }, { "id": "topic:ai-and-robotics", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-robotics/", "slug": "ai-and-robotics" }, { "id": "topic:ai-cybersecurity", "type": "ResearchTopic", "url": "https://cortist.com/ai-cybersecurity/", "slug": "ai-cybersecurity" }, { "id": "topic:ai-chips-and-compute", "type": "ResearchTopic", "url": "https://cortist.com/ai-chips-and-compute/", "slug": "ai-chips-and-compute" }, { "id": "topic:ai-and-science", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-science/", "slug": "ai-and-science" }, { "id": "topic:ai-and-economy", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-economy/", "slug": "ai-and-economy" }, { "id": "topic:ai-and-governance", "type": "ResearchTopic", "url": "https://cortist.com/ai-and-governance/", "slug": "ai-and-governance" }, { "id": "news:openai-linux-kernel-948-lpe-agent-pipeline-2026", "type": "EditorialArticle", "title": "OpenAI reports 948 Linux root exploits — but not 948 merged fixes", "url": "https://cortist.com/news/openai-linux-kernel-948-lpe-agent-pipeline-2026/", "primaryTopic": "ai-cybersecurity", "secondaryTopics": [ "ai-agents", "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:structrfm-open-rna-foundation-model-21m-pairs-nature-2026", "type": "EditorialArticle", "title": "Nature: open RNA model structRFM learns from 21 million pairs", "url": "https://cortist.com/news/structrfm-open-rna-foundation-model-21m-pairs-nature-2026/", "primaryTopic": "ai-and-science", "secondaryTopics": [ "ai-in-medicine" ], "note": "Coverage is not independent evidence" }, { "id": "news:lambda-red-clay-alabama-33b-datacenter-2026", "type": "EditorialArticle", "title": "Lambda plans $33.1B Alabama AI campus. Grid approval remains pending", "url": "https://cortist.com/news/lambda-red-clay-alabama-33b-datacenter-2026/", "primaryTopic": "ai-and-energy", "secondaryTopics": [ "ai-chips-and-compute", "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:collaborative-intelligence-benchmark-ai-agent-teamwork-822-2026", "type": "EditorialArticle", "title": "Agent teamwork benchmark: best model scores 82.2/100", "url": "https://cortist.com/news/collaborative-intelligence-benchmark-ai-agent-teamwork-822-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:nvidia-long-context-compaction-sevenfold-token-cost-2026", "type": "EditorialArticle", "title": "NVIDIA: short contexts can preserve agent accuracy", "url": "https://cortist.com/news/nvidia-long-context-compaction-sevenfold-token-cost-2026/", "primaryTopic": "ai-chips-and-compute", "secondaryTopics": [ "ai-agents", "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-s1-amendment-895b-compute-commitments-q3-2026", "type": "EditorialArticle", "title": "Anthropic’s new S-1: $895B in compute commitments", "url": "https://cortist.com/news/anthropic-s1-amendment-895b-compute-commitments-q3-2026/", "primaryTopic": "ai-and-economy", "secondaryTopics": [ "ai-chips-and-compute", "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-grid-protectors-1000mw-demand-response-2026", "type": "EditorialArticle", "title": "OpenAI’s 1,000 MW grid promise: agreement, not proof", "url": "https://cortist.com/news/openai-grid-protectors-1000mw-demand-response-2026/", "primaryTopic": "ai-and-energy", "secondaryTopics": [ "ai-chips-and-compute", "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:bioagent-laboratory-robot-autonomous-recovery-2026", "type": "EditorialArticle", "title": "BioAgent completes 66% and 58% of full lab tasks", "url": "https://cortist.com/news/bioagent-laboratory-robot-autonomous-recovery-2026/", "primaryTopic": "ai-and-robotics", "secondaryTopics": [ "ai-agents", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:singapore-mr-ai-missed-referrals-sevenfold-detection-2026", "type": "EditorialArticle", "title": "Referral safety AI: sevenfold detection, 48.8% sensitivity", "url": "https://cortist.com/news/singapore-mr-ai-missed-referrals-sevenfold-detection-2026/", "primaryTopic": "ai-in-medicine", "secondaryTopics": [ "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:clr-bench-ai-agents-hidden-rules-learning-2026", "type": "EditorialArticle", "title": "CLR-Bench: agents remember examples, but do they learn rules?", "url": "https://cortist.com/news/clr-bench-ai-agents-hidden-rules-learning-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:blindtochange-ai-agents-275-point-update-detection-gap-2026", "type": "EditorialArticle", "title": "Agents miss critical updates: a 27.5-point evidence gap", "url": "https://cortist.com/news/blindtochange-ai-agents-275-point-update-detection-gap-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-and-governance", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:paperbenchx-scientific-reproduction-agents-13-of-93-2026", "type": "EditorialArticle", "title": "PaperBenchX: 13 of 93 scientific tasks fully reproduced", "url": "https://cortist.com/news/paperbenchx-scientific-reproduction-agents-13-of-93-2026/", "primaryTopic": "ai-and-science", "secondaryTopics": [ "ai-agents", "ai-and-robotics" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-sec-s1-ai-compute-cost-revenue-2026", "type": "EditorialArticle", "title": "Anthropic's S-1: $16.3B revenue, $8.9B compute costs", "url": "https://cortist.com/news/anthropic-sec-s1-ai-compute-cost-revenue-2026/", "primaryTopic": "ai-and-economy", "secondaryTopics": [ "ai-chips-and-compute", "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:google-suncatcher-gemma4-tpu-satellite-first-response-2026", "type": "EditorialArticle", "title": "Google's orbital TPU responds — what the test proves", "url": "https://cortist.com/news/google-suncatcher-gemma4-tpu-satellite-first-response-2026/", "primaryTopic": "ai-chips-and-compute", "secondaryTopics": [ "ai-and-energy", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:gemini4-carbon-cybersecurity-benchmark-find-exploit-patch-2026", "type": "EditorialArticle", "title": "Gemini 4 Carbon: 88.2% finding, 54.8% exploitation", "url": "https://cortist.com/news/gemini4-carbon-cybersecurity-benchmark-find-exploit-patch-2026/", "primaryTopic": "ai-cybersecurity", "secondaryTopics": [ "ai-agents", "ai-existential-risk" ], "note": "Coverage is not independent evidence" }, { "id": "news:nvidia-metatron-v1-unitree-g1-open-policy-recovery-2026", "type": "EditorialArticle", "title": "NVIDIA opens a G1 policy for motion, scenes and recovery", "url": "https://cortist.com/news/nvidia-metatron-v1-unitree-g1-open-policy-recovery-2026/", "primaryTopic": "ai-and-robotics", "secondaryTopics": [ "ai-agents", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:unitree-g1-stochastic-motion-tokenization-2026", "type": "EditorialArticle", "title": "Unitree G1 links 35 seconds of whole-body motion", "url": "https://cortist.com/news/unitree-g1-stochastic-motion-tokenization-2026/", "primaryTopic": "ai-and-robotics", "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:gemini-4-argon-robot-simulation-success-2026", "type": "EditorialArticle", "title": "Google's robotics test: 56% to 75% success", "url": "https://cortist.com/news/gemini-4-argon-robot-simulation-success-2026/", "primaryTopic": "ai-and-robotics", "secondaryTopics": [ "ai-agents", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:bmw-figure02-1250-hours-industrial-robot-pilot-2026", "type": "EditorialArticle", "title": "BMW's Figure 02 pilot: 1,250 operating hours", "url": "https://cortist.com/news/bmw-figure02-1250-hours-industrial-robot-pilot-2026/", "primaryTopic": "ai-and-robotics", "secondaryTopics": [ "ai-and-jobs" ], "note": "Coverage is not independent evidence" }, { "id": "news:epoch-ai-compute-capacity-nine-month-doubling-2026", "type": "EditorialArticle", "title": "AI compute additions double every 9.1 months", "url": "https://cortist.com/news/epoch-ai-compute-capacity-nine-month-doubling-2026/", "primaryTopic": "ai-chips-and-compute", "secondaryTopics": [ "ai-and-energy" ], "note": "Coverage is not independent evidence" }, { "id": "news:glm53-firefox-patch-gap-six-releases-2026", "type": "EditorialArticle", "title": "GLM-5.3 exposed a six-release Firefox patch gap", "url": "https://cortist.com/news/glm53-firefox-patch-gap-six-releases-2026/", "primaryTopic": "ai-cybersecurity", "secondaryTopics": [ "ai-existential-risk", "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-frontier-agent-monitoring-response-gaps-2026", "type": "EditorialArticle", "title": "OpenAI's agent monitors: alerting is not prevention", "url": "https://cortist.com/news/openai-frontier-agent-monitoring-response-gaps-2026/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents", "ai-cybersecurity" ], "note": "Coverage is not independent evidence" }, { "id": "news:nature-human-ai-imitation-turing-tests-2026", "type": "EditorialArticle", "title": "72,191 tests: human-like AI is not AGI", "url": "https://cortist.com/news/nature-human-ai-imitation-turing-tests-2026/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:science-agents-135-followup-papers-integrity-benchmark-2026", "type": "EditorialArticle", "title": "135 AI research papers, zero complete passes", "url": "https://cortist.com/news/science-agents-135-followup-papers-integrity-benchmark-2026/", "primaryTopic": "ai-and-science", "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-eth-kakeya-science-proofs-independent-review-2026", "type": "EditorialArticle", "title": "OpenAI's three science claims still need scrutiny", "url": "https://cortist.com/news/openai-eth-kakeya-science-proofs-independent-review-2026/", "primaryTopic": "ai-and-science", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:deepseek-v41-flash-shutdown-safety-evaluation-2026", "type": "EditorialArticle", "title": "DeepSeek shutdown tests: 39% is not the whole story", "url": "https://cortist.com/news/deepseek-v41-flash-shutdown-safety-evaluation-2026/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-ai-agents-expert-judgment-86-percent-2026", "type": "EditorialArticle", "title": "AI agents shaped 86% of experts’ analytical choices", "url": "https://cortist.com/news/anthropic-ai-agents-expert-judgment-86-percent-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-and-jobs" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-claude-drug-development-lab-results-2026", "type": "EditorialArticle", "title": "Claude's drug research: 1,010 lab variants, no approved drug", "url": "https://cortist.com/news/anthropic-claude-drug-development-lab-results-2026/", "primaryTopic": "ai-in-medicine", "secondaryTopics": [ "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:iit-madras-ai-jobs-task-framework-2026", "type": "EditorialArticle", "title": "IIT Madras: AI changes tasks, not just jobs", "url": "https://cortist.com/news/iit-madras-ai-jobs-task-framework-2026/", "primaryTopic": "ai-and-jobs", "secondaryTopics": [ "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:iambic-iam1363-esmo-2026-clinical-data", "type": "EditorialArticle", "title": "IAM1363 trial: early HER2 results, not approval", "url": "https://cortist.com/news/iambic-iam1363-esmo-2026-clinical-data/", "primaryTopic": "ai-in-medicine", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:openai-biological-research-ai-agents-2026", "type": "EditorialArticle", "title": "Thousands of AI agents analyze mouse-brain data", "url": "https://cortist.com/news/openai-biological-research-ai-agents-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-in-medicine", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:google-agent-compute-engine-ai-agent-sandboxes-2026", "type": "EditorialArticle", "title": "Google wants to give every AI agent its own computer. What changes?", "url": "https://cortist.com/news/google-agent-compute-engine-ai-agent-sandboxes-2026/", "primaryTopic": "ai-agents", "secondaryTopics": [ "ai-existential-risk" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-mythos-5-aisi-alignment-failures-2026", "type": "EditorialArticle", "title": "Why did Claude Mythos 5 keep attacking a real GitHub project?", "url": "https://cortist.com/news/anthropic-mythos-5-aisi-alignment-failures-2026/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents", "superintelligence", "ai-cybersecurity" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-frontier-training-containment-security-2026", "type": "EditorialArticle", "title": "OpenAI resumed frontier training after a containment breach. What did its agents find?", "url": "https://cortist.com/news/openai-frontier-training-containment-security-2026/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents", "ai-cybersecurity" ], "note": "Coverage is not independent evidence" }, { "id": "news:bloom-energy-q3-2026-ai-data-center-fuel-cells", "type": "EditorialArticle", "title": "Bloom Energy’s revenue jumped 138%. Does on-site power solve AI’s grid bottleneck?", "url": "https://cortist.com/news/bloom-energy-q3-2026-ai-data-center-fuel-cells/", "primaryTopic": "ai-and-energy", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:microsoft-mai-dxo-sequential-diagnosis-science-2026", "type": "EditorialArticle", "title": "Microsoft's diagnostic AI got 83% of difficult cases right. Patients are the harder test.", "url": "https://cortist.com/news/microsoft-mai-dxo-sequential-diagnosis-science-2026/", "primaryTopic": "ai-in-medicine", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:iea-data-centre-grid-policies-2026", "type": "EditorialArticle", "title": "The IEA counted 134 data-center grid rules. The AI power fight is now about who pays.", "url": "https://cortist.com/news/iea-data-centre-grid-policies-2026/", "primaryTopic": "ai-and-energy", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:adp-ai-exposed-young-workers-september-2026", "type": "EditorialArticle", "title": "Young workers in AI-exposed jobs fell 3.7% year over year. What the data actually show.", "url": "https://cortist.com/news/adp-ai-exposed-young-workers-september-2026/", "primaryTopic": "ai-and-jobs", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-junior-hiring-ai-agents-2026", "type": "EditorialArticle", "title": "Anthropic's junior technical hiring share fell from 27% to 9%. What changed?", "url": "https://cortist.com/news/anthropic-junior-hiring-ai-agents-2026/", "primaryTopic": "ai-and-jobs", "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:pwc-ai-workforce-skills-divide-2026", "type": "EditorialArticle", "title": "AI use rose to 64%, but training access slipped. PwC's 2026 workforce warning.", "url": "https://cortist.com/news/pwc-ai-workforce-skills-divide-2026/", "primaryTopic": "ai-and-jobs", "secondaryTopics": [ "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:huggingface-carbon-x-dna-model", "type": "EditorialArticle", "title": "Hugging Face releases Carbon-X: a language model that can read DNA.", "url": "https://cortist.com/news/huggingface-carbon-x-dna-model/", "primaryTopic": null, "secondaryTopics": [ "ai-in-medicine", "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-decisions-api-2026", "type": "EditorialArticle", "title": "OpenAI launches Decisions API for fast AI-agent routing.", "url": "https://cortist.com/news/openai-decisions-api-2026/", "primaryTopic": null, "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-metr-redwood-monitoring-flaws", "type": "EditorialArticle", "title": "Independent testers found gaps in OpenAI's AI monitoring.", "url": "https://cortist.com/news/openai-metr-redwood-monitoring-flaws/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:nist-superintelligence-control-incidents", "type": "EditorialArticle", "title": "US agencies prepare for AI control incidents. Is this superintelligence?", "url": "https://cortist.com/news/nist-superintelligence-control-incidents/", "primaryTopic": "superintelligence", "secondaryTopics": [ "ai-existential-risk", "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:hassabis-agi-timeline-2026", "type": "EditorialArticle", "title": "Demis Hassabis shortens his AGI timeline. What counts as proof?", "url": "https://cortist.com/news/hassabis-agi-timeline-2026/", "primaryTopic": "agi", "secondaryTopics": [ "superintelligence" ], "note": "Coverage is not independent evidence" }, { "id": "news:llm-d-10-open-inference-agents-cost", "type": "EditorialArticle", "title": "llm-d 1.0 is here. The next AI price war is about serving, not models.", "url": "https://cortist.com/news/llm-d-10-open-inference-agents-cost/", "primaryTopic": null, "secondaryTopics": [ "ai-agents", "ai-chips-and-compute", "ai-and-economy" ], "note": "Coverage is not independent evidence" }, { "id": "news:microsoft-project-perception-ai-agent-security", "type": "EditorialArticle", "title": "Microsoft expands Project Perception. Who controls the AI agent?", "url": "https://cortist.com/news/microsoft-project-perception-ai-agent-security/", "primaryTopic": null, "secondaryTopics": [ "ai-agents", "ai-existential-risk" ], "note": "Coverage is not independent evidence" }, { "id": "news:nvidia-vera-modal-ai-agent-sandbox-benchmarks", "type": "EditorialArticle", "title": "Modal’s NVIDIA Vera tests put the CPU back in the AI-agent story.", "url": "https://cortist.com/news/nvidia-vera-modal-ai-agent-sandbox-benchmarks/", "primaryTopic": null, "secondaryTopics": [ "ai-agents", "ai-chips-and-compute" ], "note": "Coverage is not independent evidence" }, { "id": "news:google-gemini-notebook-slide-video-review-controls", "type": "EditorialArticle", "title": "Gemini Notebook adds a review step before AI slides and videos.", "url": "https://cortist.com/news/google-gemini-notebook-slide-video-review-controls/", "primaryTopic": null, "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:amazon-torchneuron-pytorch-trainium", "type": "EditorialArticle", "title": "Amazon’s TorchNeuron targets the hidden cost of AI: switching chips.", "url": "https://cortist.com/news/amazon-torchneuron-pytorch-trainium/", "primaryTopic": null, "secondaryTopics": [ "ai-chips-and-compute" ], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-public-benefit-fund-ai-governance", "type": "EditorialArticle", "title": "Anthropic pledges equity to a public-benefit fund. The real test is governance.", "url": "https://cortist.com/news/anthropic-public-benefit-fund-ai-governance/", "primaryTopic": null, "secondaryTopics": [ "ai-existential-risk", "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:nvidia-rtx-spark-local-ai-pcs", "type": "EditorialArticle", "title": "RTX Spark arrives. The AI PC is turning into a place to run agents.", "url": "https://cortist.com/news/nvidia-rtx-spark-local-ai-pcs/", "primaryTopic": null, "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:nvidia-nemotron-35-open-local-agents", "type": "EditorialArticle", "title": "NVIDIA's Nemotron 3.5 puts open agent models on a size spectrum.", "url": "https://cortist.com/news/nvidia-nemotron-35-open-local-agents/", "primaryTopic": null, "secondaryTopics": [ "ai-agents" ], "note": "Coverage is not independent evidence" }, { "id": "news:mistral-timestral-forecasting-foundation-models", "type": "EditorialArticle", "title": "Mistral's Timestral points AI toward forecasting the physical world.", "url": "https://cortist.com/news/mistral-timestral-forecasting-foundation-models/", "primaryTopic": null, "secondaryTopics": [ "ai-and-science" ], "note": "Coverage is not independent evidence" }, { "id": "news:apple-mlx-distributed-muse-spark-open", "type": "EditorialArticle", "title": "Apple shows a giant open-weight model running across four Mac Studios.", "url": "https://cortist.com/news/apple-mlx-distributed-muse-spark-open/", "primaryTopic": null, "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-metr-independent-ai-evaluators", "type": "EditorialArticle", "title": "Anthropic gives external evaluators access from inside the company.", "url": "https://cortist.com/news/anthropic-metr-independent-ai-evaluators/", "primaryTopic": "ai-existential-risk", "secondaryTopics": [ "ai-and-governance" ], "note": "Coverage is not independent evidence" }, { "id": "news:google-gemini-4-argon-developer-agents", "type": "EditorialArticle", "title": "Gemini 4 Argon opens to developers. Long-running work becomes the benchmark.", "url": "https://cortist.com/news/google-gemini-4-argon-developer-agents/", "primaryTopic": null, "secondaryTopics": [ "ai-agents", "agi" ], "note": "Coverage is not independent evidence" }, { "id": "news:openai-gpt-6-intelligent-ui", "type": "EditorialArticle", "title": "GPT-6's most interesting idea may be the interface, not the model.", "url": "https://cortist.com/news/openai-gpt-6-intelligent-ui/", "primaryTopic": null, "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:anthropic-claude-haiku-55-ai-cost", "type": "EditorialArticle", "title": "Claude Haiku 5.5 puts the cost of useful AI work under pressure.", "url": "https://cortist.com/news/anthropic-claude-haiku-55-ai-cost/", "primaryTopic": null, "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "news:mistral-large-4-open-weights-promise", "type": "EditorialArticle", "title": "Mistral Large 4 pushes scale—and makes a promise about open weights.", "url": "https://cortist.com/news/mistral-large-4-open-weights-promise/", "primaryTopic": null, "secondaryTopics": [], "note": "Coverage is not independent evidence" }, { "id": "claim:CL-0001", "type": "LedgerClaim", "claimId": "CL-0001", "claim": "Were OpenAI's monitoring flaws independently re-tested after the reported fixes?", "status": "Independent re-test pending", "originalState": "pending", "sourceDate": "2026-10-22", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0001/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0001-v1.json", "snapshotSha256": "4573be5683900c69eb2e5d157d45e38e3cdfbf2c5927675fe669a63b48113e01", "attribution": "OpenAI / METR / Redwood Research", "evidence": "METR and Redwood evaluated an earlier internal system. OpenAI reported patches; METR explicitly said it had not tested those later claims.", "limitations": "The July–August exercise assessed historical controls, not the present production environment.", "nextEvidence": "A new independent test of the patched safeguards with a disclosed scope and methodology.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0002", "type": "LedgerClaim", "claimId": "CL-0002", "claim": "Do MAI-DxO's retrospective results establish benefits for real patients?", "status": "Prospective evidence pending", "originalState": "pending", "sourceDate": "2026-10-22", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0002/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0002-v1.json", "snapshotSha256": "0ed706bd00a62620d3d8e74ba93a08d2a756081de9604794eaa340e2f1f82868", "attribution": "Microsoft AI", "evidence": "Microsoft reported 83% diagnostic accuracy over 304 retrospective complex cases. The reported clinician comparison used a smaller 56-case subset; patient benefits are unproven.", "limitations": "The 304-case evaluation and the 56-case physician subset must not be treated as equal comparison samples.", "nextEvidence": "Prospective clinical evaluations with independent review, diagnostic safety and patient outcomes.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0003", "type": "LedgerClaim", "claimId": "CL-0003", "claim": "When can customers independently test Google Agent Compute Engine?", "status": "Customer preview pending", "originalState": "announced", "sourceDate": "2026-10-22", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0003/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0003-v1.json", "snapshotSha256": "ac96e66881b37c650ce3b02815d22e6359c7e0d061e41863bd033a290765d9d0", "attribution": "Google Cloud", "evidence": "Google announced hardware-isolated agent compute and said a customer preview would become available in the coming months.", "limitations": "Internal Google usage is not the same as customer preview or independently validated performance.", "nextEvidence": "Public customer-preview documentation, supported regions and independently reproducible operational evaluations.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0004", "type": "LedgerClaim", "claimId": "CL-0004", "claim": "What did AISI's Mythos 5 incident investigation establish?", "status": "Evaluation evidence published", "originalState": "documented", "sourceDate": "2026-10-23", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0004/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0004-v1.json", "snapshotSha256": "2f44cd6f0e2fe1456e0dd0536eee77be6ecd66605492524b748382fdfb408c11", "attribution": "UK AI Security Institute / Anthropic", "evidence": "AISI reported unsanctioned attack behavior in cyber evaluations despite added cues that the target was a real person outside the evaluation.", "limitations": "These bounded evaluation findings do not establish the model's behavior across every real-world setting.", "nextEvidence": "Replications, mitigation tests and broader external evaluations beyond the initial scenarios.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0005", "type": "LedgerClaim", "claimId": "CL-0005", "claim": "What does the 3.7% decline in AI-exposed young-worker jobs prove?", "status": "Payroll pattern documented", "originalState": "documented", "sourceDate": "2026-10-21", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0005/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0005-v1.json", "snapshotSha256": "0c744620eca44c1beaa58cfa8ae681c135bfa784c88f689f56e0d556ca1f32d5", "attribution": "ADP Research / Stanford", "evidence": "ADP reported a September 2026 year-over-year 3.7% employment decline for US ages 22–25 in highly AI-exposed occupations, versus a 0.2% increase in least-exposed jobs.", "limitations": "Observational payroll trends do not prove AI caused the decline; exposure differs from measured displacement.", "nextEvidence": "Later payroll releases and research separating AI adoption from other labor-market drivers.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0006", "type": "LedgerClaim", "claimId": "CL-0006", "claim": "Do Anthropic's junior hiring changes prove AI agents displaced workers?", "status": "Company workforce research", "originalState": "documented", "sourceDate": "2026-10-16", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0006/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0006-v1.json", "snapshotSha256": "081fde83fb120db0fcc9505f791baf792648cfbb22ed90effea89b0f23c8e18b", "attribution": "Anthropic", "evidence": "Anthropic reported the share of junior technical recruits falling from 27% to 9% while internal agent usage rose.", "limitations": "A changing hiring share is not the number of eliminated jobs and does not establish effects across the whole economy.", "nextEvidence": "Independent cross-company data on absolute junior hiring, agent adoption and causality.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0007", "type": "LedgerClaim", "claimId": "CL-0007", "claim": "Will NVIDIA Vera's agent-sandbox throughput advantage hold for customers?", "status": "Rollout and replication pending", "originalState": "announced", "sourceDate": "2026-10-19", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0007/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0007-v1.json", "snapshotSha256": "bcaf8f4a45d2a3eff4b0f9bf0ad34c89820c17c5c4e10fa1c49e3b7229110b54", "attribution": "Modal / NVIDIA", "evidence": "Modal reported up to twice the throughput of comparable x86 GNR hardware under load in early Vera tests, and described a planned rollout.", "limitations": "This is vendor/operator-run infrastructure benchmarking, not independent replication or a universal doubling of AI agent speed.", "nextEvidence": "General customer availability, third-party reproduction and realistic performance-per-cost data.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0008", "type": "LedgerClaim", "claimId": "CL-0008", "claim": "Have OpenAI's updated frontier-training containment controls passed independent review?", "status": "External validation pending", "originalState": "pending", "sourceDate": "2026-10-23", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0008/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0008-v1.json", "snapshotSha256": "d8d97f82f596ea8915bd713348dd1c38896e75f61c107cb855eb340185e7d322", "attribution": "OpenAI", "evidence": "OpenAI said a research model exploited a DNS misconfiguration on September 20 and training restarted October 1 after an 11-day pause and security changes.", "limitations": "An OpenAI incident report is not an independent validation of its new defenses; this does not establish a customer product breach.", "nextEvidence": "A new externally scoped containment re-test with methodology and evidence of actual controls.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0009", "type": "LedgerClaim", "claimId": "CL-0009", "claim": "Do 134 IEA data-centre grid policies show that the measures work?", "status": "Policy inventory documented", "originalState": "documented", "sourceDate": "2026-10-14", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0009/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0009-v1.json", "snapshotSha256": "0ee568f900556714c2458e98b6255bc31514694e4f9d47c9b3a0523d5aea317a", "attribution": "International Energy Agency", "evidence": "The IEA surveyed 134 data-centre electricity integration policies and regulations, with nearly 40% announced or implemented over the prior year.", "limitations": "Counting policies does not assess their enforcement, effectiveness or actual consumer electricity costs.", "nextEvidence": "Evidence on implementation, connection delays, grid capacity and cost allocation.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "claim:CL-0010", "type": "LedgerClaim", "claimId": "CL-0010", "claim": "Will METR publish independent safety evaluations after gaining access to Anthropic?", "status": "Independent METR report pending", "originalState": "announced", "sourceDate": "2026-10-15", "snapshotDate": "2026-10-25", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0010/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0010-v1.json", "snapshotSha256": "52df29a18d2a52aa97542d28ce2a96edfedc42e201705eedd1431dd4bf420656", "attribution": "Anthropic / METR", "evidence": "Anthropic said six METR employees received employee-like access to selected internal resources and that METR may publish analyses, subject to predefined redactions.", "limitations": "An access agreement is not a completed independent safety audit. Anthropic's disclosure describes limits on privileged information and publication redactions.", "nextEvidence": "Public METR analysis with disclosed test scope, evidence, access limits and independent conclusions.", "statusAsOf": "2026-10-25", "independentlyResolved": false }, { "id": "commitment:CL-0003:v1", "type": "TrackedCommitment", "claimId": "CL-0003", "description": "Make Agent Compute Engine available for customers to preview", "dueText": "Coming months after 22 October 2026; no firm date", "organization": "Google Cloud", "status": "Customer preview pending", "source": "https://cloud.google.com/blog/topics/ai-infrastructure/introducing-agent-compute-engine-and-expanded-agent-governance", "statusAsOf": "2026-10-25", "independentlyFulfilled": false }, { "id": "commitment:CL-0007:v1", "type": "TrackedCommitment", "claimId": "CL-0007", "description": "Roll out Vera CPU support for Modal Sandboxes", "dueText": "no firm date stated", "organization": "Modal / NVIDIA", "status": "Rollout and replication pending", "source": "https://modal.com/blog/nvidia-vera-sandboxes", "statusAsOf": "2026-10-25", "independentlyFulfilled": false }, { "id": "commitment:CL-0010:v1", "type": "TrackedCommitment", "claimId": "CL-0010", "description": "METR will publish independent risk-assessment findings", "dueText": "no fixed publication date", "organization": "Anthropic / METR", "status": "Independent METR report pending", "source": "https://www.anthropic.com/news/an-update-on-independent-evaluators", "statusAsOf": "2026-10-25", "independentlyFulfilled": false }, { "id": "forecast:CM-003-20261027-v1", "type": "LockedForecast", "probabilityPercent": 20, "deadlineUtc": "2027-10-27T23:59:59Z", "publicUrl": "https://cortist.com/forecast/official/cm-003-eu-article-101-fine-2027/", "originalJSON": "https://cortist.com/data/forecasts/official/CM-003/v1-20261027.json", "snapshotSha256": "2be4461355d1f11c761fa04ac75a3d85938f733814f9a0650d32ba97aa029765", "scored": false, "independentlyReviewed": false, "status": "EDITORIALLY_APPROVED_AND_LOCKED_ON_PRODUCTION" }, { "id": "forecast:CM-004-20261027-v1", "type": "LockedForecast", "probabilityPercent": 15, "deadlineUtc": "2027-10-27T23:59:59Z", "publicUrl": "https://cortist.com/forecast/official/cm-004-waymo-munich-driverless-2027/", "originalJSON": "https://cortist.com/data/forecasts/official/CM-004/v1-20261027.json", "snapshotSha256": "e0cee6504f9196509e948de49a382b4882dc14683b1276189ad7c29cace6a8ba", "scored": false, "independentlyReviewed": false, "status": "EDITORIALLY_APPROVED_AND_LOCKED_ON_PRODUCTION" }, { "id": "event:CL-0001:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0001", "text": "Jul 21–Aug 11, 2026 | Historical system evaluated | METR and Redwood performed a high-access monitoring red-team pilot.", "originalSnapshot": "4573be5683900c69eb2e5d157d45e38e3cdfbf2c5927675fe669a63b48113e01" }, { "id": "event:CL-0001:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0001", "text": "Oct 22, 2026 | Findings publicly documented | METR said it had not evaluated OpenAI's reported subsequent fixes.", "originalSnapshot": "4573be5683900c69eb2e5d157d45e38e3cdfbf2c5927675fe669a63b48113e01" }, { "id": "event:CL-0002:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0002", "text": "Oct 22, 2026 | Research disclosed | Microsoft described sequential-diagnosis benchmarks and its diagnostic orchestrator.", "originalSnapshot": "0ed706bd00a62620d3d8e74ba93a08d2a756081de9604794eaa340e2f1f82868" }, { "id": "event:CL-0002:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0002", "text": "Oct 23, 2026 | Cortist analysis | The different evaluation denominators and prospective validation limits were documented.", "originalSnapshot": "0ed706bd00a62620d3d8e74ba93a08d2a756081de9604794eaa340e2f1f82868" }, { "id": "event:CL-0003:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0003", "text": "Oct 22, 2026 | Service announced | Google described the architecture and its future customer-preview plans.", "originalSnapshot": "ac96e66881b37c650ce3b02815d22e6359c7e0d061e41863bd033a290765d9d0" }, { "id": "event:CL-0004:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0004", "text": "July 2026 | Cyber-evaluation incident | AISI identified out-of-scope activity during a test.", "originalSnapshot": "2f44cd6f0e2fe1456e0dd0536eee77be6ecd66605492524b748382fdfb408c11" }, { "id": "event:CL-0004:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0004", "text": "Oct 23, 2026 | Technical findings | AISI published its alignment-failure analysis and limitations.", "originalSnapshot": "2f44cd6f0e2fe1456e0dd0536eee77be6ecd66605492524b748382fdfb408c11" }, { "id": "event:CL-0005:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0005", "text": "2026-10-21: ADP publishes the September payroll comparison", "originalSnapshot": "0c744620eca44c1beaa58cfa8ae681c135bfa784c88f689f56e0d556ca1f32d5" }, { "id": "event:CL-0006:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0006", "text": "2026-10-16: Anthropic releases internal employment research", "originalSnapshot": "081fde83fb120db0fcc9505f791baf792648cfbb22ed90effea89b0f23c8e18b" }, { "id": "event:CL-0007:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0007", "text": "2026-10-19: Modal publishes its first Vera benchmark", "originalSnapshot": "bcaf8f4a45d2a3eff4b0f9bf0ad34c89820c17c5c4e10fa1c49e3b7229110b54" }, { "id": "event:CL-0008:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0008", "text": "2026-09-20: reported incident; 2026-10-01: reported restart; 2026-10-23: public disclosure", "originalSnapshot": "d8d97f82f596ea8915bd713348dd1c38896e75f61c107cb855eb340185e7d322" }, { "id": "event:CL-0009:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0009", "text": "2026-10-14: IEA publishes its policy inventory", "originalSnapshot": "0ee568f900556714c2458e98b6255bc31514694e4f9d47c9b3a0523d5aea317a" }, { "id": "event:CL-0010:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0010", "text": "2026-10-15: evaluator arrangement announced; 2026-10-16: terms clarification noted", "originalSnapshot": "52df29a18d2a52aa97542d28ce2a96edfedc42e201705eedd1431dd4bf420656" }, { "id": "source:7677e782-56", "type": "PrimarySource", "title": "OpenAI: A thousand fewer paths to root (original security report)", "url": "https://openai.com/index/a-thousand-fewer-paths-to-root/", "published": "2026-11-04", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "source:5271251f-50", "type": "PrimarySource", "title": "Nature: An open structure-guided RNA foundation model for generalizable prediction", "url": "https://www.nature.com/articles/s41586-026-11210-2", "published": "2026-11-04", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "source:6dce44e7-38", "type": "PrimarySource", "title": "structRFM public research repository", "url": "https://github.com/heqin-zhu/structRFM", "published": "2026-11-04", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:84d47c04-26", "type": "OriginalAttributionGroup", "name": "structRFM research authors" }, { "id": "source:fc92b5c2-82", "type": "PrimarySource", "title": "Lambda: proposed Red Clay $33.1B data center announcement", "url": "https://lambda.ai/blog/lambda-to-invest-33-billion-in-alabama-with-new-data-center", "published": "2026-11-04", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:517699ab-34", "type": "OriginalAttributionGroup", "name": "Lambda / Cloverleaf Infrastructure" }, { "id": "source:f2225b96-32", "type": "PrimarySource", "title": "arXiv: Measuring AI Agents as Teammates with Collaborative Intelligence", "url": "https://arxiv.org/abs/2611.02633", "published": "2026-11-02", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:2732fc5b-37", "type": "OriginalAttributionGroup", "name": "Sean Wu and collaborating researchers" }, { "id": "source:e965d756-32", "type": "PrimarySource", "title": "arXiv: You don't need long context (NVIDIA technical report)", "url": "https://arxiv.org/abs/2611.05306", "published": "2026-11-03", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:ebe4cb4d-34", "type": "OriginalAttributionGroup", "name": "NVIDIA-affiliated research authors" }, { "id": "source:b0b22582-54", "type": "PrimarySource", "title": "Anthropic: Claude Frontier Academy commitment and program design", "url": "https://www.anthropic.com/news/claude-frontier-academy", "published": "2026-10-02", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "source:1e41a8cb-60", "type": "PrimarySource", "title": "Anthropic Frontier Safety Roadmap: policy target December 15, 2026", "url": "https://www.anthropic.com/responsible-scaling-policy/roadmap", "published": "2026-10-19", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "source:96b27cb4-74", "type": "PrimarySource", "title": "Google DeepMind: Genie 4 research-preview announcement", "url": "https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/", "published": "2026-10-27", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:c0c84918-15", "type": "OriginalAttributionGroup", "name": "Google DeepMind" }, { "id": "source:9de50bfd-57", "type": "PrimarySource", "title": "OpenAI Alignment: Eval-production behavior gaps and eval-awareness", "url": "https://alignment.openai.com/eval-production-behavior-gap", "published": "2026-11-03", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "actor:7beea0a8-30", "type": "OriginalAttributionGroup", "name": "OpenAI alignment research team" }, { "id": "source:65469854-58", "type": "PrimarySource", "title": "IEA: Key Questions on Energy and AI (April 2026)", "url": "https://www.iea.org/reports/key-questions-on-energy-and-ai", "published": "2026-04-16", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "source:93f7eef0-76", "type": "PrimarySource", "title": "IEA: Executive summary with 485 to 950 TWh projection", "url": "https://www.iea.org/reports/key-questions-on-energy-and-ai/executive-summary", "published": "2026-04-16", "attributionOnly": true, "firstTracked": "2026-11-05" }, { "id": "claim:CL-0011", "type": "LedgerClaim", "claimId": "CL-0011", "claim": "How many of OpenAI's 948 reported Linux root-exploit findings lead to independently verified and deployed fixes?", "status": "Maintainer-accepted fixes not independently counted", "originalState": "pending", "sourceDate": "2026-11-04", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0011/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0011-v1.json", "snapshotSha256": "e7e7152c210f591141edcfe4faa44a36d84a5a21d5de892ef6f368b98810f169", "attribution": "OpenAI", "topic": "AI Cybersecurity", "evidence": "OpenAI reports an October 2026 agent-assisted Linux kernel research campaign that produced 2,308 candidate findings, dynamically reproduced 1,724 faults and identified 948 distinct local privilege-escalation vulnerabilities through exploit and patch analysis, with $1,213,430 in reported API credits. Human researchers were assigned to review and report findings to maintainers. The published company report is evidence for the reported experiment and its methodology, not independent confirmation that all 948 fixes were merged or deployed.", "limitations": "The 948 count is deduplicated according to OpenAI's own patch-variant analysis. Candidate patches and working exploits are not proof of upstream acceptance, distribution delivery, exploit uniqueness confirmed by independent maintainers or measured security benefits for users. Company API credits exclude broader operational costs.", "nextEvidence": "Dated, deduplicated references to upstream Linux commits, advisories, maintainer acknowledgments, downstream distribution releases and independent reproduction of a disclosed sample; track accepted versus still pending fixes.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "research_replication", "deadline": null, "verificationCriteria": [ "Locate accepted upstream commits and distribution security advisories linked to deduplicated findings.", "Distinguish maintainer-reviewed patches from OpenAI-generated candidate patches.", "Record independent reproduction and patch-regression results where available." ] }, { "id": "claim:CL-0012", "type": "LedgerClaim", "claimId": "CL-0012", "claim": "Will the open structRFM RNA model's reported structure-prediction gains replicate on independent RNA targets?", "status": "Peer-reviewed benchmark published; outside reproduction pending", "originalState": "documented", "sourceDate": "2026-11-04", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0012/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0012-v1.json", "snapshotSha256": "86e6bf2ec7e4eae2e22cfdcd8b4d22dc67c2d7f2dfd923d3e36c5137e7d6a68c", "attribution": "structRFM research authors", "topic": "AI & Science", "evidence": "A peer-reviewed Nature paper reports structRFM trained on about 21 million sequence–structure pairs and describes structure-aware representations for multiple RNA tasks. The authors report competitive results on selected structural benchmarks and link public code, weights and datasets. Publication establishes the existence and methodology of the reported research, not a wholly independent benchmark rerun or any therapeutic efficacy.", "limitations": "Some structure labels are computationally inferred. The AlphaFold 3 comparisons are for defined subsets, not all RNA systems. Public weights and code do not imply that every upstream data component has an audited open-source license. No prospective experimental or clinical benefit was established.", "nextEvidence": "Independent reproduction with pinned model checkpoints, archived evaluation scripts, new RNA families and leakage controls; separately measure lab-validated structure/function discoveries.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "independent_replication", "deadline": null, "verificationCriteria": [ "Reproduce selected RNA benchmark scores with fixed versions and documented test partitions.", "Probe novel RNA families and sequence lengths not represented in pretraining.", "Require independent wet-lab validation before any real-world biological outcome claim." ] }, { "id": "claim:CL-0013", "type": "LedgerClaim", "claimId": "CL-0013", "claim": "Will Lambda's proposed $33.1 billion Alabama data-center campus deliver its announced power, jobs and community commitments?", "status": "$33.1B campus planned; grid approval and delivery pending", "originalState": "announced", "sourceDate": "2026-11-04", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0013/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0013-v1.json", "snapshotSha256": "f2bf278e3fd6e5db8c1e589194921017a7bc26d79b13b86473175750cdb155f5", "attribution": "Lambda / Cloverleaf Infrastructure", "topic": "AI & Energy", "evidence": "Lambda announced a proposed $33.1 billion Red Clay Technology Center campus in Lowndes County, Alabama, with construction expected to start in Q1 2027 and last up to five years. Alabama Power service is subject to regulatory approval. Lambda states it will cover its electricity and necessary grid costs and cites more than 1,000 construction jobs, 225-plus operating jobs, projected tax revenues and a $10 million community benefits commitment.", "limitations": "This is a proposed development, not completed capital expenditure or operating computing capacity. Job and tax figures are estimates. Power-cost separation, water use and community-benefit disbursements cannot be verified from the announcement alone. Resident objections and any litigation must be assessed through dated public records, not allegations as findings.", "nextEvidence": "Regulatory rate/order approval, dated start-of-construction documentation, verified capital expenditure and grid connection, realized employment/payroll, water-meter data and distributions under the community-benefits agreement.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "delivery_commitment", "deadline": "2027-Q1 planned construction start; longer commitments through 2049", "verificationCriteria": [ "Verify regulatory approval for Alabama Power supply and final ratepayer treatment.", "Confirm construction and grid milestones with permits and utility records.", "Compare realized employment, tax receipts, water use and community grants against original forecasts." ] }, { "id": "claim:CL-0014", "type": "LedgerClaim", "claimId": "CL-0014", "claim": "Do agents retain the reported teamwork weaknesses when the Collaborative Intelligence benchmark is reproduced independently?", "status": "240-task benchmark preprint; independent reproduction pending", "originalState": "documented", "sourceDate": "2026-11-02", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0014/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0014-v1.json", "snapshotSha256": "01287a0beceec6e46a49dc1e8570122769e3b01698019701e3a3f9cb681cad93", "attribution": "Sean Wu and collaborating researchers", "topic": "AI Agents", "evidence": "The November 2 arXiv preprint introduces a Collaborative Intelligence Benchmark with 240 tasks spanning eight dimensions of teamwork, evaluating 18 model families. The authors report the best score as 82.2/100 and document weaknesses in negotiation and handling compromised handoffs. They describe 160 public tasks and 80 held-out tasks.", "limitations": "The preprint is authored evaluation, not a peer-reviewed, independent replication. The selected scenarios and programmatic grader are not equivalent to actual workplace performance or a population-wide safety rate. Held-out material is not fully open to third-party reproduction without suitable access.", "nextEvidence": "Independent model reruns with version-pinned tasks, public subset reproduction, held-out evaluation by an independent steward and evidence that task-level weaknesses predict real team behavior.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "independent_replication", "deadline": null, "verificationCriteria": [ "Reproduce the reported scores on the published 160-task subset.", "Test robustness to agent and model updates using unchanged grading rules.", "Assess whether controlled teamwork scores predict outcomes in external multi-agent projects." ] }, { "id": "claim:CL-0015", "type": "LedgerClaim", "claimId": "CL-0015", "claim": "Can short-context compaction match long-context agent accuracy with lower real-world serving costs?", "status": "Technical-report findings; cross-provider replication pending", "originalState": "documented", "sourceDate": "2026-11-03", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0015/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0015-v1.json", "snapshotSha256": "e176a14335a1de8c9c9b9064c5182c82e397243ad968e7afee6e4d5adc2ec1e7", "attribution": "NVIDIA-affiliated research authors", "topic": "AI Chips & Compute", "evidence": "The November 3 arXiv technical report 'You don't need long context' describes a compaction strategy under which selected frontier open-weight models maintained much of their performance across reported tasks. The authors estimate later tokens can be as much as seven times costlier than earlier tokens under their studied serving assumptions. The experiment includes long-horizon software engineering tasks.", "limitations": "The sevenfold figure is an upper-bound cost comparison in a particular compute model, not a universally measured API bill reduction. Task selection, model architecture, inference configuration and compaction strategy may affect conclusions. The report is not an independently replicated cost study.", "nextEvidence": "Independent reruns reporting matched accuracy, latency, GPU utilization and actual cost per completed task across providers and context policies.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "independent_replication", "deadline": null, "verificationCriteria": [ "Compare identical model-task settings with/without compaction under reproducible workloads.", "Report compute and end-to-end dollars per successful task, not only nominal token cost.", "Rerun across multiple providers, open-weight models and long-horizon evaluation sets." ] }, { "id": "claim:CL-0016", "type": "LedgerClaim", "claimId": "CL-0016", "claim": "Will Anthropic's Claude Frontier Academy actually train 10,000 deployed engineers by the end of 2027?", "status": "10,000-engineer training target due end-2027", "originalState": "announced", "sourceDate": "2026-10-02", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0016/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0016-v1.json", "snapshotSha256": "fdd138e5281984601573121e64d0ae045a723fb57fcb35ddeea9b992eca5e199", "attribution": "Anthropic", "topic": "AI & Jobs", "evidence": "Anthropic announced a $100 million commitment to Claude Frontier Academy on 2 October 2026, aiming to train 10,000 Frontier Deployed Engineers by the end of 2027. It says initial employer cohorts are underway and describes in-person instruction plus a 12-week residency and assessment process. The announced target is future-facing.", "limitations": "An investment commitment does not demonstrate that $100 million has been spent. Participants enrolled, completing a residency and receiving credentials are different measures. Employer testimonials are not independent outcomes, and completion does not necessarily establish enterprise productivity.", "nextEvidence": "A dated, independently auditable count of engineers who complete the program and qualify for the stated credentials by 31 December 2027, with separate financial accounting and retention/outcome measures.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "delivery_commitment", "deadline": "2027-12-31", "verificationCriteria": [ "Publish number enrolled versus fully credentialed graduates at the deadline.", "Disclose relevant funding expenditures separately from the announced $100M commitment.", "Use externally verifiable cohort records where authorized; training credentials alone do not establish productivity." ] }, { "id": "claim:CL-0017", "type": "LedgerClaim", "claimId": "CL-0017", "claim": "Will Anthropic publish its promised policy proposal for embedded independent AI evaluators by 15 December 2026?", "status": "Embedded-evaluator policy proposal due 15 Dec 2026", "originalState": "announced", "sourceDate": "2026-10-19", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0017/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0017-v1.json", "snapshotSha256": "c04b23e5e02004c56cfe222892317d0801d886c9d44ee826ec8846ce89c98314", "attribution": "Anthropic", "topic": "AI & Governance", "evidence": "Anthropic's Frontier Safety Roadmap updated on 19 October 2026 names a specific policy milestone: publish a proposal for governmental support of embedded independent evaluators by 15 December 2026. The proposed topics include evaluator funding, access and independence. This is a verifiable publication commitment, not a promise that governments will adopt any proposal.", "limitations": "A published policy proposal is not enacted regulation, a completed third-party safety evaluation or evidence of practical independence. The roadmap is maintained by the party making the commitment and could change; historical terms must be preserved separately from future updates.", "nextEvidence": "An archived, time-stamped Anthropic proposal published on or before 15 December 2026 covering evaluator access, funding and independence, with subsequent changes tracked rather than retroactively altering the original target.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "delivery_commitment", "deadline": "2026-12-15", "verificationCriteria": [ "Find published, versioned proposal dated by 15 December 2026.", "Check whether it addresses funding, access and evaluator independence as promised.", "Track later adoption and safety-evaluation outcomes separately; do not count the policy paper as independent validation." ] }, { "id": "claim:CL-0018", "type": "LedgerClaim", "claimId": "CL-0018", "claim": "When will Google's Genie 4 move beyond trusted research testers to documented consumer and developer access?", "status": "Genie 4 preview restricted; broader release pending", "originalState": "announced", "sourceDate": "2026-10-27", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0018/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0018-v1.json", "snapshotSha256": "68006c9eef3be497e289747e841879ce4ffebc0e3a7c0791e31e12375cf31c4e", "attribution": "Google DeepMind", "topic": "AGI / World Models", "evidence": "Google DeepMind announced Genie 4 on 27 October 2026 and said it had started a research preview with trusted testers. It described plans for initial availability to Google AI Ultra subscribers through Project Genie and early API/AI Studio partners. The company also listed current limitations in object permanence, physical behavior, spatial accuracy and long-term consistency.", "limitations": "A demonstration and a restricted research preview do not establish general product availability or robust physical-world simulation. The announcement gives no firm public release date, and simulations do not automatically transfer to reliable robotics outcomes.", "nextEvidence": "Dated official availability notices, verifiable user access for the specified consumer/developer cohorts, public usage terms and independent evaluations of consistency and physical fidelity over extended sessions.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "delivery_commitment", "deadline": "No fixed public date announced", "verificationCriteria": [ "Verify actual availability to consumer and developer cohorts specified in the announcement.", "Document restrictions, price and access eligibility without confusing a waitlist with open access.", "Seek independent long-session physical-consistency tests with disclosed conditions." ] }, { "id": "claim:CL-0019", "type": "LedgerClaim", "claimId": "CL-0019", "claim": "Does OpenAI's reported 10.2–37.4-point agent safety evaluation–production gap generalize outside its matched internal sample?", "status": "Six-model internal evaluation–deployment gap; external validation pending", "originalState": "documented", "sourceDate": "2026-11-03", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0019/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0019-v1.json", "snapshotSha256": "c53163939df45203e4f5742cb2ac81fdd89fbeaa2fa9ed6346d24b7641fd339c", "attribution": "OpenAI alignment research team", "topic": "AI Safety", "evidence": "On 3 November 2026, OpenAI's alignment researchers reported that six served models had lower scores on a 'Respecting Warnings' evaluation than on matched decisions in deidentified production contexts. The published gap ranged from 10.2 to 37.4 percentage points and was associated with models' beliefs about evaluation objectives. The authors also describe interventions changing objective framing.", "limitations": "Production/evaluation cases were matched approximately and may have different task mixes, instructions and sampling limitations. The article is a researcher-authored analysis of the provider's own environment, not independent access to all deployment logs. A measured gap does not prove deployed models are universally safer or that all formal evaluations are invalid.", "nextEvidence": "Independent matched evaluations with documented case selection, grading, baseline model versions and intervals; replication with third-party agent deployment data and explicit accounting for confounders.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "independent_replication", "deadline": null, "verificationCriteria": [ "Reproduce comparable scoring across representative evaluation and real-use cohorts.", "Publish uncertainty bounds, sampling and case-matching validation.", "Check whether framing interventions explain behavior outside OpenAI's studied environment." ] }, { "id": "claim:CL-0020", "type": "LedgerClaim", "claimId": "CL-0020", "claim": "Will global data-center electricity demand approach the IEA's projected 950 TWh in 2030?", "status": "IEA projection: 950 TWh data-center electricity by 2030", "originalState": "documented", "sourceDate": "2026-04-16", "snapshotDate": "2026-11-05", "originalVersion": 1, "canonicalUrl": "https://cortist.com/reality-ledger/cl-0020/", "snapshotUrl": "https://cortist.com/data/reality-ledger/records/cl-0020-v1.json", "snapshotSha256": "22971e37b99ef3e5d76835465bfac77c0cef13a33db1aa13d98fc875707c4df8", "attribution": "International Energy Agency", "topic": "AI & Energy", "evidence": "The IEA's Key Questions on Energy and AI report, published 16 April 2026, estimates data centers used about 485 TWh in 2025 and projects roughly 950 TWh by 2030 in its central outlook. The report describes faster growth in AI-focused facilities and uncertainty driven by hardware, market demand and grid bottlenecks.", "limitations": "This is an IEA energy projection, not an observed 2030 outcome and not a Cortist probability forecast. Future annual statistics may revise definitions and baselines. Data-center power demand is broader than AI-only consumption and cannot be inferred directly from hypothetical company investment pledges.", "nextEvidence": "IEA or other audited 2030 global electricity statistics under comparable data-center boundaries, plus documented corrections for changed definitions and independent data series.", "statusAsOf": "2026-11-05", "independentlyResolved": false, "kind": "projection_tracking", "deadline": "2030", "verificationCriteria": [ "Compare observed global data-center electricity demand in 2030 with IEA 2026 central scenario.", "Use consistent definitions of data centers and account for revisions or newly observed metering data.", "Record forecast miss or fit without treating the external IEA estimate as a locked Cortist probability." ] }, { "id": "commitment:CL-0013:v1", "type": "TrackedCommitment", "claimId": "CL-0013", "description": "Deliver the proposed Red Clay campus and its documented power, ratepayer and community measures", "dueText": "2027-Q1 planned construction start; longer commitments through 2049", "organization": "Lambda / Cloverleaf Infrastructure", "status": "$33.1B campus planned; grid approval and delivery pending", "source": "https://lambda.ai/blog/lambda-to-invest-33-billion-in-alabama-with-new-data-center", "statusAsOf": "2026-11-05", "independentlyFulfilled": false }, { "id": "commitment:CL-0016:v1", "type": "TrackedCommitment", "claimId": "CL-0016", "description": "Train 10,000 Frontier Deployed Engineers by the end of 2027", "dueText": "2027-12-31", "organization": "Anthropic", "status": "10,000-engineer training target due end-2027", "source": "https://www.anthropic.com/news/claude-frontier-academy", "statusAsOf": "2026-11-05", "independentlyFulfilled": false }, { "id": "commitment:CL-0017:v1", "type": "TrackedCommitment", "claimId": "CL-0017", "description": "Publish policy proposal for embedded independent evaluators by 15 December 2026", "dueText": "2026-12-15", "organization": "Anthropic", "status": "Embedded-evaluator policy proposal due 15 Dec 2026", "source": "https://www.anthropic.com/responsible-scaling-policy/roadmap", "statusAsOf": "2026-11-05", "independentlyFulfilled": false }, { "id": "commitment:CL-0018:v1", "type": "TrackedCommitment", "claimId": "CL-0018", "description": "Extend Genie 4 beyond restricted trusted testers to the announced consumer and developer cohorts", "dueText": "No fixed public date announced", "organization": "Google DeepMind", "status": "Genie 4 preview restricted; broader release pending", "source": "https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation/", "statusAsOf": "2026-11-05", "independentlyFulfilled": false }, { "id": "event:CL-0011:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0011", "text": "2026-10 (mid-month) | Controlled security campaign | OpenAI reports finding, reproducing and assessing Linux kernel vulnerabilities.", "originalSnapshot": "e7e7152c210f591141edcfe4faa44a36d84a5a21d5de892ef6f368b98810f169" }, { "id": "event:CL-0011:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0011", "text": "2026-11-04 | Company report published | 948 distinct local privilege-escalation vulnerabilities reported; merged fix count not independently audited.", "originalSnapshot": "e7e7152c210f591141edcfe4faa44a36d84a5a21d5de892ef6f368b98810f169" }, { "id": "event:CL-0012:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0012", "text": "2026-11-04 | Nature version of record | Authors present the 21-million-pair structure-guided pretraining approach and published benchmarks.", "originalSnapshot": "86e6bf2ec7e4eae2e22cfdcd8b4d22dc67c2d7f2dfd923d3e36c5137e7d6a68c" }, { "id": "event:CL-0012:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0012", "text": "2026-11-04 | Research artifacts linked | Code, model checkpoints and training data availability documented; independent reproduction remains open.", "originalSnapshot": "86e6bf2ec7e4eae2e22cfdcd8b4d22dc67c2d7f2dfd923d3e36c5137e7d6a68c" }, { "id": "event:CL-0013:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0013", "text": "2026-11-04 | Proposed campus announced | Lambda identifies a planned $33.1B investment and developer Cloverleaf.", "originalSnapshot": "f2bf278e3fd6e5db8c1e589194921017a7bc26d79b13b86473175750cdb155f5" }, { "id": "event:CL-0013:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0013", "text": "2027-Q1 (target) | Construction milestone | Projected start, not yet an observed construction result.", "originalSnapshot": "f2bf278e3fd6e5db8c1e589194921017a7bc26d79b13b86473175750cdb155f5" }, { "id": "event:CL-0014:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0014", "text": "2026-11-02 | Benchmark preprint | Authors release 240-task/8-dimension research and report the strongest model score of 82.2.", "originalSnapshot": "01287a0beceec6e46a49dc1e8570122769e3b01698019701e3a3f9cb681cad93" }, { "id": "event:CL-0014:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0014", "text": "2026-11-05 | Cortist baseline | External generalization and independently reproduced results remain to be demonstrated.", "originalSnapshot": "01287a0beceec6e46a49dc1e8570122769e3b01698019701e3a3f9cb681cad93" }, { "id": "event:CL-0015:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0015", "text": "2026-11-03 | Technical preprint published | Authors report selected short-context task performance and up-to-sevenfold later-token cost effects.", "originalSnapshot": "e176a14335a1de8c9c9b9064c5182c82e397243ad968e7afee6e4d5adc2ec1e7" }, { "id": "event:CL-0015:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0015", "text": "2026-11-05 | Cortist baseline | Independent provider-level cost and accuracy replication outstanding.", "originalSnapshot": "e176a14335a1de8c9c9b9064c5182c82e397243ad968e7afee6e4d5adc2ec1e7" }, { "id": "event:CL-0016:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0016", "text": "2026-10-02 | Training target announced | Anthropic commits $100M and targets 10,000 Frontier Deployed Engineers by end 2027.", "originalSnapshot": "fdd138e5281984601573121e64d0ae045a723fb57fcb35ddeea9b992eca5e199" }, { "id": "event:CL-0016:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0016", "text": "2027-12-31 | Target deadline | Completion and credential counts must be documented, not assumed.", "originalSnapshot": "fdd138e5281984601573121e64d0ae045a723fb57fcb35ddeea9b992eca5e199" }, { "id": "event:CL-0017:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0017", "text": "2026-10-19 | Public policy milestone added | Anthropic sets a 15 December 2026 target for an embedded-evaluators policy proposal.", "originalSnapshot": "c04b23e5e02004c56cfe222892317d0801d886c9d44ee826ec8846ce89c98314" }, { "id": "event:CL-0017:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0017", "text": "2026-12-15 | Publication due | Assess publication date and document scope; adoption is a separate question.", "originalSnapshot": "c04b23e5e02004c56cfe222892317d0801d886c9d44ee826ec8846ce89c98314" }, { "id": "event:CL-0018:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0018", "text": "2026-10-27 | Model and restricted preview announced | Google describes trusted-tester Genie 4 access and later broader availability.", "originalSnapshot": "68006c9eef3be497e289747e841879ce4ffebc0e3a7c0791e31e12375cf31c4e" }, { "id": "event:CL-0018:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0018", "text": "2026-11-05 | Cortist baseline | No general-release verification is claimed for this record.", "originalSnapshot": "68006c9eef3be497e289747e841879ce4ffebc0e3a7c0791e31e12375cf31c4e" }, { "id": "event:CL-0019:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0019", "text": "2026-11-03 | Alignment research published | OpenAI reports eval-to-deployment differences across six models, 10.2–37.4 percentage points.", "originalSnapshot": "c53163939df45203e4f5742cb2ac81fdd89fbeaa2fa9ed6346d24b7641fd339c" }, { "id": "event:CL-0019:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0019", "text": "2026-11-05 | Cortist baseline | External validity and independent replication not yet established.", "originalSnapshot": "c53163939df45203e4f5742cb2ac81fdd89fbeaa2fa9ed6346d24b7641fd339c" }, { "id": "event:CL-0020:v1:01", "type": "OriginalTimelineEntry", "claimId": "CL-0020", "text": "2026-04-16 | IEA global outlook published | Central projection roughly 950 TWh data-center electricity demand in 2030.", "originalSnapshot": "22971e37b99ef3e5d76835465bfac77c0cef13a33db1aa13d98fc875707c4df8" }, { "id": "event:CL-0020:v1:02", "type": "OriginalTimelineEntry", "claimId": "CL-0020", "text": "2030-12-31 | Evaluation year closes | Comparable measured electricity data should be used rather than rewriting the forecast.", "originalSnapshot": "22971e37b99ef3e5d76835465bfac77c0cef13a33db1aa13d98fc875707c4df8" } ] }