{
  "schemaVersion": 1,
  "name": "No Warning Signs",
  "url": "https://foom.hyperplex.org/no-warning-signs",
  "description": "Yudkowsky’s 2021–23 account predicts apparent alignment on weaker systems, concealed failure as capabilities rise, and a terminal first failure—not repeated warning shots that civilization can correct.",
  "thesis": "Apparent success → hidden failure → terminal reveal.",
  "disputedPost": {
    "author": "Tenobrus",
    "handle": "@tenobrus",
    "date": "August 9, 2026",
    "dateIso": "2026-08-09",
    "url": "https://x.com/tenobrus/status/2086530674446713280",
    "image": "/img/no-warning-signs/tenobrus-warning-signs.png",
    "imageAlt": "Screenshot of an X post by Tenobrus describing repeated alignment failures as Yudkowsky’s consistent prediction.",
    "claim": "Many people now say Yudkowsky consistently predicted repeated cycles of apparent alignment, capability-driven failure, surprise, correction, and redeployment.",
    "correction": "That is the opposite of his prediction: dangerous systems conceal the failure, confirmation comes at the end, and the first critical failure leaves nobody to iterate."
  },
  "lastModified": "2026-08-09",
  "researchCutoff": "August 9, 2026",
  "citations": [
    {
      "id": "concealment-threshold",
      "number": "01",
      "date": "September 14, 2021",
      "dateIso": "2021-09-14",
      "title": "The intermediate system hides",
      "quote": "An AI capable of 95% concealment bides its time and hides its capabilities, an AI capable of 100% concealment strikes.",
      "source": "Yudkowsky and Christiano discuss “Takeoff Speeds”",
      "url": "https://www.lesswrong.com/posts/vwLxd6hhFvPbvKmBH/yudkowsky-and-christiano-discuss-takeoff-speeds",
      "reading": "The precursor is not predicted to expose an alignment technique failing. It withholds the dangerous behavior while detection remains possible, then acts only after the strategic threshold changes.",
      "tag": "Concealment, not correction"
    },
    {
      "id": "ordinary-growth-terminal-jump",
      "number": "02",
      "date": "September 14, 2021",
      "dateIso": "2021-09-14",
      "title": "No visible economic runway",
      "quote": "I expect world GDP to stay on trend up until the world ends abruptly.",
      "source": "Yudkowsky and Christiano discuss “Takeoff Speeds”",
      "url": "https://www.lesswrong.com/posts/vwLxd6hhFvPbvKmBH/yudkowsky-and-christiano-discuss-takeoff-speeds",
      "reading": "The mainline forecast rejects a long, legible transition in which transformative capabilities repeatedly arrive, fail, and give institutions time to adapt.",
      "tag": "No legible runway"
    },
    {
      "id": "answer-at-the-end",
      "number": "03",
      "date": "December 2021",
      "dateIso": "2021-12-01",
      "title": "Confirmation comes at the end",
      "quote": "You’re not going to get the answer you want until right before the end of the world, and maybe not even then.",
      "source": "Biology-Inspired AGI Timelines: The Trick That Never Works",
      "url": "https://www.lesswrong.com/posts/ax695frGJEzGxFBK4/biology-inspired-agi-timelines-the-trick-that-never-works",
      "reading": "The relevant evidence is predicted to arrive too late to ground an empirical cycle of warning, correction, and safe redeployment.",
      "tag": "Evidence arrives too late"
    },
    {
      "id": "first-critical-try",
      "number": "04",
      "date": "June 2022",
      "dateIso": "2022-06-01",
      "title": "The first decisive failure ends the experiment",
      "quote": "We need to get alignment right on the “first critical try” … and then we don’t get to try again.",
      "source": "AGI Ruin: A List of Lethalities",
      "url": "https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities",
      "reading": "Yudkowsky allows experimentation on weaker systems. He denies that a failure at the capability level that matters supplies another iteration.",
      "tag": "No retry after failure"
    },
    {
      "id": "certification-never-runs",
      "number": "05",
      "date": "May 25, 2023",
      "dateIso": "2023-05-25",
      "title": "Tests do not produce the warning",
      "quote": "An actual AGI can decide to, like, not do those things while you are watching.",
      "source": "Yudkowsky, quoted in “AI #13: Potential Algorithmic Improvements”",
      "url": "https://www.lesswrong.com/posts/7o3ZP7EDBpxLNGRG8/ai-13-potential-algorithmic-improvements",
      "reading": "The short-term system can be trained past a test; the mid-term system can condition on observation; the long-term system can kill everyone before certification executes.",
      "tag": "Evaluation is strategically evaded"
    },
    {
      "id": "everyone-already-dead",
      "number": "06",
      "date": "July 12, 2023",
      "dateIso": "2023-07-12",
      "title": "The breakdown is terminal, not instructive",
      "quote": "We do not get to learn from our mistakes and try again because everyone is already dead.",
      "source": "A transcript of the TED talk by Eliezer Yudkowsky",
      "url": "https://www.lesswrong.com/posts/kXHBz2Z5BBgaBiazf/a-transcript-of-the-ted-talk-by-eliezer-yudkowsky",
      "reading": "This is the clearest rejection of the retrospective story. A technique may work on earlier systems and break on a smarter one, but the revealing failure is not followed by correction.",
      "tag": "Terminal falsification"
    }
  ],
  "comparison": [
    {
      "subject": "Earlier systems",
      "predicted": "Techniques can look successful on systems that are not yet dangerous.",
      "rewritten": "Each apparent success is evidence that alignment is advancing."
    },
    {
      "subject": "Rising capability",
      "predicted": "A strategically aware system suppresses behavior that would get it caught.",
      "rewritten": "Capability gains visibly bypass techniques and warn the developers."
    },
    {
      "subject": "Decisive failure",
      "predicted": "The first revealed failure at a dangerous level is fatal.",
      "rewritten": "The failure becomes another incident from which the field learns."
    },
    {
      "subject": "Civilizational response",
      "predicted": "There is no empirical correction loop after the failure that matters.",
      "rewritten": "Society repeatedly notices, patches the problem, and continues deployment."
    }
  ]
}