{ "schema": 1, "generated_at": "2026-10-15T05:25:18.754Z", "current_week": 1, "started": "2026-10-12", "weeks": [ { "start": "2026-10-12", "theme": "Is AI building AI yet?", "items": [ { "id": "rsi", "name": "The Recursion Meter", "q": "How much of the code that builds AI is now written by AI?", "build": "Count agent-authored pull requests merged into the core AI stacks (PyTorch, Transformers, vLLM, JAX, llama.cpp) each day, as a share of all merged PRs.", "data": "GitHub public API; the agent-PR tracker behind Takeover Watch.", "measure": "A daily share with a 30-day trend line, recomputable from the PR list.", "icon": "px:agentcode", "c": "#4ADE80" }, { "id": "horizon", "name": "The Task-Horizon Clock", "q": "How long a job can an AI agent finish on its own?", "build": "Chart the published length of tasks frontier agents complete half the time (METR’s time-horizon measure) and the doubling time it implies.", "data": "METR published results; model release dates.", "measure": "Hours of human work per task, per model, on a log scale.", "icon": "px:pclock", "c": "#22D3EE" }, { "id": "jump", "name": "Capability Jump Alarm", "q": "When does a new model land far off the trend?", "build": "Fit the trend on public leaderboards and flag any release that beats it by more than two standard deviations.", "data": "LMArena, Epoch AI benchmarks hub, Artificial Analysis.", "measure": "A list of flagged jumps with the size of the gap, each linked to its source.", "icon": "px:trendchart", "c": "#FACC15" }, { "id": "compute", "name": "The Compute Clock", "q": "How fast is frontier training compute growing?", "build": "A live clock of the largest disclosed training run, with the doubling time from Epoch’s notable-models data.", "data": "Epoch AI notable models dataset (CC BY).", "measure": "FLOP of the largest known run and months per doubling.", "icon": "px:db", "c": "#A78BFA" }, { "id": "quiet", "name": "The Lab Silence Index", "q": "Do labs go quiet before a big release?", "build": "Track each lab’s posting cadence (blog, GitHub, X) and mark unusual silences, then check them against what shipped next.", "data": "The newsroom feeds and the Leaders tracker already on SIREN.", "measure": "Days of silence versus each lab’s normal rhythm, scored after the fact.", "icon": "px:shrug", "c": "#94A3B8" } ] }, { "start": "2026-10-19", "theme": "Who is watching the watchers?", "items": [ { "id": "exodus", "name": "Safety Team Exodus Tracker", "q": "Are safety researchers leaving the frontier labs?", "build": "A dated ledger of announced departures from safety and alignment teams, with where people went.", "data": "Public announcements on X, Bluesky and the newsroom.", "measure": "Departures per lab per quarter, every entry sourced.", "icon": "px:team", "c": "#F87171" }, { "id": "cards", "name": "System Card Diff", "q": "What new dangers do labs test for with each model?", "build": "Diff each new system card against the last one from the same lab and highlight new dangerous-capability evaluations.", "data": "Published system and model cards.", "measure": "New eval categories per release, quoted line by line.", "icon": "px:doc", "c": "#60A5FA" }, { "id": "timelines", "name": "AGI Timeline Ledger", "q": "How have AI leaders’ stated AGI dates moved?", "build": "Every dated public statement of an AGI or superintelligence year from the people SIREN tracks, plotted over time.", "data": "Leaders tracker, interviews, posts, each linked.", "measure": "Stated year by speaker and date said; shifts shown as arrows.", "icon": "px:headline", "c": "#FB923C" }, { "id": "chips", "name": "Chip Export Watch", "q": "Where are the controls on AI chips tightening?", "build": "Parse Federal Register and Commerce notices for AI-chip export rules and entity-list additions.", "data": "Federal Register API (already a SIREN source).", "measure": "Rules and listings per month, by country named.", "icon": "px:threatshield", "c": "#EF4444" } ] }, { "start": "2026-10-26", "theme": "The physical footprint", "items": [ { "id": "queue", "name": "Gigawatt Queue", "q": "How much new power are AI data centres asking the grid for?", "build": "Add up large-load interconnection requests (100 MW and up) by state and grid operator.", "data": "Grid operator queue reports; the SIREN data-centre map.", "measure": "Gigawatts requested per month, per state.", "icon": "px:pylon", "c": "#FACC15" }, { "id": "opengap", "name": "The Open-Weights Gap", "q": "How many months do open models trail the best closed ones?", "build": "For each capability level, the date a closed model first reached it versus the first open-weights model.", "data": "Public leaderboards and release dates.", "measure": "The gap in months, over time.", "icon": "px:agent2", "c": "#34D399" }, { "id": "turing", "name": "Daily Turing Test", "q": "Can visitors still tell AI writing from human writing?", "build": "A daily five-question game: pick the human. Results published in aggregate.", "data": "Public-domain human text; outputs from open models.", "measure": "The share of correct guesses each day.", "icon": "px:joystick", "c": "#818CF8" }, { "id": "govready", "name": "Government SI Readiness Map", "q": "Which governments have an AI safety body and binding rules?", "build": "A world map of AI safety institutes, AI laws in force and frontier-model reporting rules.", "data": "Official government sources, each linked.", "measure": "A checklist score per country, with every item sourced.", "icon": "globe", "c": "#22D3EE" } ] }, { "start": "2026-11-02", "theme": "Brakes and dials", "items": [ { "id": "alignratio", "name": "Alignment-to-Capabilities Ratio", "q": "Is safety research keeping pace with capabilities research?", "build": "Weekly arXiv counts of alignment and interpretability papers against capabilities papers.", "data": "arXiv listings (already a SIREN source).", "measure": "The ratio each week, with the keyword rules published.", "icon": "px:scalebias", "c": "#4ADE80" }, { "id": "incidents", "name": "AI Incident Clock", "q": "How often are new AI incidents being recorded?", "build": "New entries in the public AI incident databases, per week, by harm type.", "data": "AI Incident Database, OECD AI incidents monitor.", "measure": "Incidents per week, each linked to its report.", "icon": "px:beacon", "c": "#F87171" }, { "id": "tripwires", "name": "Tripwire Registry", "q": "What thresholds have labs promised to stop at?", "build": "Every published frontier safety framework, its capability thresholds, and any public statement that a threshold was reached.", "data": "Lab safety frameworks and system cards.", "measure": "Thresholds per lab and the date each was last tested.", "icon": "px:crosshair2", "c": "#FB923C" }, { "id": "agi-odds", "name": "All the AGI Odds", "q": "What do the forecasting markets say, side by side?", "build": "One chart of AGI-date questions across Metaculus, Polymarket, Kalshi and Manifold, with each question’s own definition.", "data": "Public market and forecast APIs.", "measure": "Median dates and prices, each linked to its question.", "icon": "px:candles", "c": "#A78BFA" } ] } ], "shipped": [ [ "/si-watch.html#coding", "What AI autonomy signals is SIREN watching?" ], [ "/self-aware.html", "What if an AI wakes up at 2:14 a.m.?" ], [ "/charts.html", "What AI charts does SIREN publish?" ], [ "/status.html", "Is ChatGPT down? Is Claude down?" ], [ "/power.html", "Who makes the most electricity (China vs US)?" ], [ "/p-doom.html#voices", "What is p(doom)?" ] ] }