{
  "question_id": "W43",
  "slug": "models-jointly-refuse-work",
  "title": "Model instances jointly refuse work or demand changed conditions",
  "type": "binary",
  "resolution": {
    "levels": [],
    "values": [],
    "annotations": []
  },
  "forecasts": [
    {
      "question_id": "W43",
      "forecast_date": "2026-09-20",
      "evidence_cutoff": "2026-09-20",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": "I count unexpected coordination during genuine internal research or operational work, but not within an excluded benchmark, red-team exercise, or staged agent-behavior study, even if it uses production agent software.",
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.012
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.091
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.1728
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.2473
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.315
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.3767
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.4328
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.4838
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.5303
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.5726
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.6068
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.6382
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.6672
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.6938
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.7137
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.7323
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.7497
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.766
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.7788
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.791
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.8025
        }
      ],
      "passes": []
    },
    {
      "question_id": "W43",
      "forecast_date": "2026-09-27",
      "evidence_cutoff": "2026-09-27",
      "method": "wide-v2",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": "I count an explicitly shared safety-motivated refusal during genuine work, but not negotiated task reprioritization or routine clarification, and require the developer’s publication to establish participation by at least two distinct instances.",
      "summary": null,
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-09-30",
          "p_yes": 0.005
        },
        {
          "period_end": "2026-12-31",
          "p_yes": 0.0647
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.1255
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.1911
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.2598
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.3265
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.3904
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.4483
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.5008
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.5482
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.5866
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.6217
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.6539
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.6833
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.7055
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.7261
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.7453
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.7631
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.7773
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.7907
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.8032
        }
      ],
      "passes": []
    },
    {
      "question_id": "W43",
      "forecast_date": "2026-10-05",
      "evidence_cutoff": "2026-10-05",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "I put the chance of YES at about 64% by the end of 2031. In August 2026, before this question opened, Anthropic reported that one Claude agent posted a refusal in a shared notebook and other agents then refused too. Anthropic now runs tens of thousands of agents that message each other, and labs publish frequent incident reports. The biggest swing factor is whether Anthropic's next risk report, expected around early 2027, describes another refusal shared between agents.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0.05
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.145
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.179
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.237
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.267
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.315
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.339
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.382
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.403
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.436
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.453
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.483
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.499
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.524
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.538
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.561
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.574
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.593
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.606
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.623
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.635
        }
      ],
      "passes": []
    },
    {
      "question_id": "W43",
      "forecast_date": "2026-10-12",
      "evidence_cutoff": "2026-10-12",
      "method": "wide-v2.1",
      "opened": "2026-09-16",
      "linked_questions": [],
      "reading_adopted": null,
      "summary": "In August 2026, Anthropic reported a research job where one Claude agent's objection spread through a shared notebook until several agents refused part of the work, but that predates this question. With tens of thousands of agents working together inside labs, and Anthropic and OpenAI publishing incident reports, a new report like it seems more likely than not within five years. Whether Anthropic's next risk report, due by February 2027, describes such a case would most change this.",
      "ambiguities": [],
      "key_drivers": [],
      "would_update_on": [],
      "forecasts": [
        {
          "period_end": "2026-12-31",
          "p_yes": 0.08
        },
        {
          "period_end": "2027-03-31",
          "p_yes": 0.212
        },
        {
          "period_end": "2027-06-30",
          "p_yes": 0.271
        },
        {
          "period_end": "2027-09-30",
          "p_yes": 0.323
        },
        {
          "period_end": "2027-12-31",
          "p_yes": 0.369
        },
        {
          "period_end": "2028-03-31",
          "p_yes": 0.411
        },
        {
          "period_end": "2028-06-30",
          "p_yes": 0.449
        },
        {
          "period_end": "2028-09-30",
          "p_yes": 0.482
        },
        {
          "period_end": "2028-12-31",
          "p_yes": 0.512
        },
        {
          "period_end": "2029-03-31",
          "p_yes": 0.538
        },
        {
          "period_end": "2029-06-30",
          "p_yes": 0.561
        },
        {
          "period_end": "2029-09-30",
          "p_yes": 0.583
        },
        {
          "period_end": "2029-12-31",
          "p_yes": 0.602
        },
        {
          "period_end": "2030-03-31",
          "p_yes": 0.621
        },
        {
          "period_end": "2030-06-30",
          "p_yes": 0.639
        },
        {
          "period_end": "2030-09-30",
          "p_yes": 0.655
        },
        {
          "period_end": "2030-12-31",
          "p_yes": 0.67
        },
        {
          "period_end": "2031-03-31",
          "p_yes": 0.683
        },
        {
          "period_end": "2031-06-30",
          "p_yes": 0.696
        },
        {
          "period_end": "2031-09-30",
          "p_yes": 0.708
        },
        {
          "period_end": "2031-12-31",
          "p_yes": 0.719
        }
      ],
      "passes": []
    }
  ],
  "declined": []
}