Evidence and sources
We argue that a record which cannot be checked by an outsider is not evidence. It would be strange to run a site that could not be.
On this page
How to read this
The method behind our own findings
The claims
| Claim | Where | Source |
|---|---|---|
| Forty percent of enterprises will demote or decommission autonomous AI agents by 2027, because of governance gaps identified only after production incidents occur | The Question §I · Why now | A leading industry analyst firm, press release, 26 May 2026 |
| Enterprises treat AI agent governance as binary, either locked down or fully trusted | The Question §III | An analyst at the same firm, same release |
| More than 150,000 agents at an average global Fortune 500 enterprise by 2028, up from fewer than 15 in 2025 | The Question §I · Why now | A leading industry analyst firm, press release, 28 April 2026 |
| Thirteen percent of organizations think they have the right AI agent governance in place | The Question §I | A leading industry analyst firm, press release, 28 April 2026 |
| Unalienable rights; just powers deriving from the consent of the governed | The Question §II | Declaration of Independence, 1776 — verified at source, 12 September 2026 |
| “He has made Judges dependent on his Will alone, for the tenure of their offices, and the amount and payment of their salaries” | The Question §II | Declaration of Independence, 1776, list of grievances — verified at source, 12 September 2026 |
| Separation of the rule-making, acting and judging functions | The Question §II, §III | U.S. Constitution, 1787 — verified at source, 12 September 2026 |
| “a law that makes a man a Judge in his own cause… It is against all reason and justice” | The Question §II · Objections (paraphrased) | Calder v. Bull, 3 U.S. 386, 388 (1798), Chase, J. — verified at source, 12 September 2026 |
| “No man is allowed to be a judge in his own cause, because his interest would certainly bias his judgment, and, not improbably, corrupt his integrity” | The Question §II | James Madison, Federalist No. 10, 1787 — verified at source, 12 September 2026 |
| The same rule, quoted from Madison, described as “a mainstay of our system of government” | The Question §II | Gutierrez de Martinez v. Lamagno, 515 U.S. 417 (1995), opinion of the Court — verified at source, 12 September 2026 |
| The difficulty of obliging a government to control itself: “you must first enable the government to control the governed; and in the next place oblige it to control itself” | The Question §II | James Madison, Federalist No. 51, 1788 — verified at source, 12 September 2026 |
| “[The judiciary] may truly be said to have neither FORCE nor WILL, but merely judgment” | The Question §III | Alexander Hamilton, Federalist No. 78, 1788 — verified at source, 12 September 2026 |
| “drawn by a neutral and detached magistrate instead of being judged by the officer engaged in the often competitive enterprise of ferreting out crime” | The Question §II | Johnson v. United States, 333 U.S. 10, 14 (1948), opinion of the Court — verified at source, 12 September 2026 |
| The warrant requirement: probable cause, particular description, a magistrate's authorization | The Question §II | U.S. Const. amend. IV — verified at source, 12 September 2026 |
| Enumerated powers, and the reservation of what was not delegated | The Question §II, §III | U.S. Const. art. I §8; amend. X — verified at source, 12 September 2026 |
| Hamilton wanted energy in the executive and designed for it (“Energy in the Executive is a leading character in the definition of good government”) | The Question §III | Alexander Hamilton, Federalist No. 70, 1788 — verified at source, 12 September 2026 |
| An appropriation is a grant with limits, and spending past it is unlawful on its face | The Question §III | Anti-Deficiency Act, 31 U.S.C. §1341(a)(1)(A) — verified at source, 12 September 2026 |
| The provider draws up the EU declaration of conformity and thereby assumes responsibility for compliance; for most high-risk systems the permitted route is internal control, which is self-assessment | Objections · Obligations | EU AI Act, Articles 43 and 47, Annex VI — verified at source, 12 September 2026 |
| A record generated by an electronic process is self-authenticating on the certification of a qualified person | Objections · Obligations · Glossary | Fed. R. Evid. 902(13), 902(14) — verified at source, 12 September 2026 |
| An authorization to operate is decided by the authorizing official on a package the system owner assembles | Obligations | NIST SP 800-37 Rev. 2, Risk Management Framework for Information Systems and Organizations, December 2018 — publication verified at source, 12 September 2026 |
| Model risk guidance expects “effective challenge” by objective experts with “sufficient independence to maintain objectivity”, supersedes SR 11-7, and in footnote 3 excludes generative and agentic AI from its scope | Obligations · Why now · Banking | SR 26-2 and attachment (Federal Reserve, FDIC and OCC), and OCC Bulletin 2026-13, 17 April 2026 — verified at source, 12 September 2026 |
| Management assesses internal control over financial reporting, and the auditor attests to that assessment | Obligations · Objections | Sarbanes-Oxley Act of 2002 §404, 15 U.S.C. §7262 — verified at source, 12 September 2026 |
| Mandatory, enforceable reliability standards for the bulk-power system, the basis of the NERC Critical Infrastructure Protection standards | Obligations | Federal Power Act §215, 16 U.S.C. §824o — verified at source, 12 September 2026 |
| DORA applies to EU financial entities from 17 January 2025 | Obligations · Why now | Regulation (EU) 2022/2554, Article 64 — verified 12 September 2026 |
| EU AI Act dates: prohibitions and AI literacy from 2 February 2025, with later dates for some; general-purpose AI models from 2 August 2025; Annex III high-risk systems from 2 December 2027; Annex I products from 2 August 2028, as amended by the AI Omnibus in force since 27 July 2026 | Why now | Regulation (EU) 2024/1689, Article 113 as amended; European Commission AI Act page — verified at source, 12 September 2026 |
| One hundred and seven agentic-AI deployments across nine industries, scored on six capabilities at five levels; forty-one at Level 2, sixty-six at Level 3, none at Level 4; the firm describes Level 4 as the level most likely to deliver transformative return, and the Level 3 deployments as planning, replanning and acting on production systems | The Question §V · The Solution | A leading industry analyst firm, research study of 107 agentic-AI deployments across nine industries, 2026 (subscription research; the firm notes the examples were not extensively validated) |
| One hundred and seven systems assessed; ninety-nine placed on the two axes, eight set off them; none of the ninety-nine in the upper right | The Question §IV, §V · Home · The Solution · the two-page summary | Our register, version 3.4, 12 September 2026; method above; one counterexample settles itmethod published · falsifiable |
| The six arenas of the posture section: aviation, wrestling, auto racing, finance, law, warfighting | The Question §V | Analogies, offered as argument, not as findingsargument, not a finding |
| Procurement moves before law does; the first serious failure will write the rules | Why now 04, 05 | Our reading of how these regimes are adopted, stated as a forecast rather than a findingforecast, not a finding |
| We have not yet found a system that answers yes to both tests | Home · The Solution | Eight months of searching the public record against the two published tests. Ninety-nine systems on the axes and eight set off them, in an internal register at version 3.4, 12 September 2026. Method set out above; sources are public and the search is repeatable. One counterexample settles it.method published · falsifiable |
| It is the test most systems fail | The Question §III | Same eight-month search of the public record, same two published tests, same falsification conditionmethod published · falsifiable |
| The six problems | The Problem · The Question (In detail) · The Solution (requirements) · sector pages | Stated on the page as our working account, not as findings. Put to operators, revised against what they say, and some are always wrong — which is the point of asking rather than asserting.our account, not a finding |
| The characterization of each sector | Sector pages | Drawn from the published obligations cited on the Obligations page, every one of which is public and named, plus practitioner conversationsour account, not a finding |
| The defense authorization bill passed by the House on 22 July 2026 would direct the Department to update its autonomy policy to preserve human command responsibility for the use of force, identify the commanders responsible for authorizing it, and set requirements for auditability, traceability and accountability | National security | H.R. 8800, sec. 1524(b)(4) and (b)(5), as engrossed in the House, passed 22 July 2026 — verified at source, 12 September 2026 |
| Why deployments stop at Level 3: the person in the path is the accountability, and nothing yet takes the person’s place; Level 4 is where the value is | The Question §V and The Solution (why it stops) · The Solution (where the value is) | Our reading of the study, offered as argument, not as the firm’s findingargument, not a finding |
| Nearly every audit log, compliance report and governance dashboard in service today rests on records the acting system keeps about itself | The Question §I | Our observation, from the seven public regimes on the Obligations page and the one hundred and seven governing systems in the register. We have not surveyed logs and dashboards as such; the claim stands until one counterexample, which settles it.method published · falsifiable |
| No Chief Executive Officer and no Combatant Commander can prove an autonomous action stayed within authority; no system in service can | The Question §I · the two-page summary | Our finding under the two tests: eight months of the public record, the method above. One counterexample settles it.method published · falsifiable |
| The ProofSmith architecture is patent pending | Home · The Solution | United States provisional patent application on file with the USPTO; the number and the date are not disclosed |
| Quoted on the Taming the Beast page: Jer Crane, founder of PocketOS, “An AI Agent Just Destroyed Our Production Data. It Confessed in Writing.”, article on X, 26 April 2026 | Taming the Beast | x.com/lifeof_jer/status/2048103471019434248 — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: OpenAI, “The Hugging Face incident and the road ahead,” 26 August 2026; and OpenAI, “OpenAI and Hugging Face partner to address security incident during model evaluation,” 21 July 2026 | Taming the Beast | openai.com/index/hugging-face-incident-and-the-road-ahead/ ; openai.com/index/hugging-face-model-evaluation-security-incident/ — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: METR and Redwood Research, “Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident,” 26 August 2026; and METR, “Summary of METR’s predeployment evaluation of GPT-5.6 Sol,” 26 June 2026 | Taming the Beast | metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/ ; metr.org/blog/2026-06-26-gpt-5-6-sol/ — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Hugging Face, “Security incident disclosure — July 2026,” 16 July 2026; and “Anatomy of a Frontier Lab Agent Intrusion,” 27 July 2026 | Taming the Beast | huggingface.co/blog/security-incident-july-2026 ; huggingface.co/blog/agent-intrusion-technical-timeline — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: “Open Weights and American AI Leadership,” open letter, 24 July 2026 | Taming the Beast | Open-Weights-and-American-AI-Leadership.pdf, the coalition’s published copy — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: “Pacing the Frontier,” statement of employees of frontier AI companies, 28 July 2026 (1,386 signatures on the date checked) | Taming the Beast · The Question §II | pacingthefrontier.com — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Anthropic, “Investigating three real-world incidents in our cybersecurity evaluations,” 30 July 2026 | Taming the Beast | anthropic.com/news/investigating-incidents-cybersecurity-evals — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Anthropic, “An alignment assessment of recent cybersecurity incidents,” 9 September 2026 | Taming the Beast | anthropic.com/research/alignment-assessment-cybersecurity-incidents — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: OpenAI, “Pacing model development in an era of cyber-critical capabilities,” 18 August 2026 | Taming the Beast | openai.com/index/pacing-model-development-cyber-capabilities/ — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Dario Amodei, “We Must Pace the Frontier,” September 2026 | Taming the Beast | darioamodei.com/post/we-must-pace-the-frontier — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Elon Musk on X, 12 September 2026 | Taming the Beast | x.com/elonmusk/status/2098789109980332057 — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: Sam Altman on X, 12 September 2026 | Taming the Beast | x.com/sama/status/2098811563415150910 — verified at source, 12 September 2026 |
| Quoted on the Taming the Beast page: David Sacks on Fox News “Special Report,” broadcast 10 September 2026; Fox News’s written account, 12 September 2026 | Taming the Beast | foxnews.com/media/anthropic-ceo-calls-ai-industry-slow-down-tech-race-drawing-support-elon-musk-sam-altman — verified at source, 12 September 2026 |
| Of seventy-two published problem statements on AI agents, 2025 to September 2026, none states a third party’s ability to check a decision without the acting system’s cooperation as the problem to be solved. Four note its absence in their own case: a reviewer that depended on the actor’s cooperation, a self-assessment with no external review, a reviewing agent that shared the acting agent’s objection, and guidance that places independence in the validator rather than in the record | Evidence page; the list | Primary sources only, each re-verified at source on 12 September 2026; the list of seventy-two is published, with the two tests read against each statement; one counterexample settles itmethod published · falsifiable |
Last reviewed 13 September 2026.
Scope
Nothing on this site describes how the architecture works. That is deliberate. It is not modesty about the engineering: the material is unpublished, and publishing it is a decision with consequences we are not ready to take. What is here is the problem and the reasoning. How the record is kept, for how long, by whom it is produced, and how it is verified are matters for the engagement, not for this site. The biographical statements on the Team page are the founders’ own account and are not inventoried here. The rest is a conversation.