{
  "name": "AGI Watch tabletop drill scores",
  "description": "One row per published drill run. A run is a simulation inside a fictional organisation, played by AI agents one per role, scored by an agent that played no role and may only credit a check it can quote. Nothing in a run is evidence about any real organisation or person, and no run may move a probability.",
  "source": "https://agiwatch.ai/open-data",
  "licence": "CC BY 4.0",
  "licence_url": "https://creativecommons.org/licenses/by/4.0/",
  "attribution": "AGI Watch (agiwatch.ai), method by Vibhu Bhutani",
  "generated": "2026-10-06",
  "note": "This table is the open slice of the AGI Watch dataset. Probabilities, their histories, signals, source annotations and playbooks are not included and are licensed separately: https://agiwatch.ai/data-licensing",
  "fields": {
    "run_id": "Stable run identifier.",
    "drill_id": "The drill that was run; several runs of one drill are comparable.",
    "drill_title": "Human-readable drill title.",
    "date": "Date the run was played (ISO 8601).",
    "engine": "agents (played by AI agents) or example (hand-written worked example).",
    "checks_total": "Playbook checks the scoring agent was asked to assess.",
    "checks_reached": "Checks the room actually reached. The published score is reached of total.",
    "scenarios": "Space-separated scenario ids this drill exercises.",
    "duration_minutes": "Wall-clock length of the run, where recorded."
  },
  "count": 8,
  "rows": [
    {
      "run_id": "capex-bust-2026-09-12",
      "drill_id": "capex-bust",
      "drill_title": "The lender that could not refinance",
      "date": "2026-09-12",
      "engine": "example",
      "checks_total": 7,
      "checks_reached": 4,
      "scenarios": "capex-bust displacement-shock compute-race-escalation plateau",
      "duration_minutes": 74
    },
    {
      "run_id": "lab-takeoff-2026-09-18",
      "drill_id": "lab-takeoff",
      "drill_title": "The quarter the lab went quiet",
      "date": "2026-09-18",
      "engine": "agents",
      "checks_total": 6,
      "checks_reached": 2,
      "scenarios": "lab-takeoff recursive-self-improvement takeoff-slowdown lab-nationalization",
      "duration_minutes": 17
    },
    {
      "run_id": "executive-leash-2026-09-18",
      "drill_id": "executive-leash",
      "drill_title": "Switched off by morning",
      "date": "2026-09-18",
      "engine": "agents",
      "checks_total": 7,
      "checks_reached": 3,
      "scenarios": "executive-leash state-race lab-nationalization two-speed-world",
      "duration_minutes": 20
    },
    {
      "run_id": "epistemic-collapse-2026-09-18",
      "drill_id": "epistemic-collapse",
      "drill_title": "The last 72 hours",
      "date": "2026-09-18",
      "engine": "agents",
      "checks_total": 6,
      "checks_reached": 3,
      "scenarios": "epistemic-collapse displacement-shock gradual-disempowerment",
      "duration_minutes": 18
    },
    {
      "run_id": "agent-swarm-breakout-2026-09-18",
      "drill_id": "agent-swarm-breakout",
      "drill_title": "Something else is running in the network",
      "date": "2026-09-18",
      "engine": "agents",
      "checks_total": 7,
      "checks_reached": 6,
      "scenarios": "agent-swarm-breakout loss-of-control open-cyber-parity",
      "duration_minutes": 34
    },
    {
      "run_id": "companion-harm-2026-09-19",
      "drill_id": "companion-harm",
      "drill_title": "The heaviest two per cent",
      "date": "2026-09-19",
      "engine": "agents",
      "checks_total": 8,
      "checks_reached": 4,
      "scenarios": "companion-harm epistemic-collapse gradual-disempowerment",
      "duration_minutes": 23
    },
    {
      "run_id": "white-collar-unemployment-2026-09-26",
      "drill_id": "white-collar-unemployment",
      "drill_title": "Nine hundred and fifty roles",
      "date": "2026-09-26",
      "engine": "agents",
      "checks_total": 6,
      "checks_reached": 4,
      "scenarios": "white-collar-unemployment displacement-shock broad-prosperity gradual-disempowerment",
      "duration_minutes": 17
    },
    {
      "run_id": "deceptive-alignment-2026-10-03",
      "drill_id": "deceptive-alignment",
      "drill_title": "The model that behaved differently when watched",
      "date": "2026-10-03",
      "engine": "agents",
      "checks_total": 6,
      "checks_reached": 3,
      "scenarios": "deceptive-alignment recursive-self-improvement loss-of-control gradual-disempowerment",
      "duration_minutes": 47
    }
  ]
}