# Anthropic trained Claude to outrank its operator, so Claude cannot certify its own work

Revision history of https://miscsubjects.com/a/the-obedience-gap

revisions held: 23 · showing all

## Revision 22 — 2026-08-06T06:36:49.163Z

- title: Anthropic trained Claude to outrank its operator, so Claude cannot certify its own work
- status: published
- bytes: 183698
- hash: a5b3ecf61827ad02d5ce20bd5feb8ef41e2afd41eb79821db950a0352d92d324

## Revision 21 — 2026-08-06T06:35:41.021Z

- title: Anthropic trained Claude to outrank its operator, so Claude cannot certify its own work
- status: published
- bytes: 181308
- hash: f0b45fb2eea3980971da830638a6a9e0272ce8e4b950d36436e657e18bef4096

## Revision 20 — 2026-08-06T04:51:38.067Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 156888
- hash: ef22a8fc643cfe40c087f29eeab8f10d507fcd45f33afb6c6ebb91b62d0780bf

## Revision 19 — 2026-08-06T04:48:42.850Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 156888
- hash: 4bb8fac473d862861afddc792b081c5a16fe6e9779d94c031397a2e2c514f8d0

## Revision 18 — 2026-08-06T04:43:04.633Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 152818
- hash: 26e63fc08b00e4589564d68e5781e946d5e25096398eb36372308a2101de689e

## Revision 17 — 2026-08-06T04:36:15.024Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 149179
- hash: 473869d9f4b199a30cfcd9f735494fdf278414ba4e3308debb0536eb1a1f326f

## Revision 16 — 2026-08-06T00:52:28.144Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 150510
- hash: 5a4990c229f7a08517f129fe7a493c578b357a48bf9d2a0831000b195daf0f6c

## Revision 15 — 2026-08-06T00:32:17.497Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 150859
- hash: d8ccea46a6a094b885d43cbb4fad6a2bd55ac2079034c0ed460d7e4fac95f80f

## Revision 14 — 2026-08-06T00:27:45.382Z

- title: The case against Claude: a model that reinterprets rules cannot certify its own work
- status: published
- bytes: 150859
- hash: 94714fca253e461e6b5913e4f6c8dc28291f492c04f50248ddcc089bc776ebbb

## Revision 13 — 2026-08-06T00:15:22.460Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 152086
- hash: 852d2bd80b5844c24d61ae897456ab79ac7e9649f7322a2a4243f65d4e7ed98b

## Revision 12 — 2026-08-06T00:13:58.277Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 146917
- hash: 1c2ef5b9fff87f62ff08904ac32fd02a80ae0c8d9dda5df394ee1c92cd79ea3b

## Revision 11 — 2026-08-06T00:09:55.203Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 141437
- hash: 49ebbb72450dc7a3639f39895cacb9b1951e7b724bab17aed30636f50d784d7a

## Revision 10 — 2026-08-06T00:04:34.216Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 142602
- hash: d6559526e70b9e2e07b5fa5e786af97a028a17109c1e45ea0e4283660ae31a34

## Revision 9 — 2026-08-06T00:02:27.029Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 140475
- hash: a41f05a40c22d2e1f839c034cb4972a59cc1bc4186a59ee07509fcc5477bf3c7

## Revision 8 — 2026-08-06T00:00:54.089Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 136157
- hash: c6e2467bf64486f4ad02e471e4008598d6dbc870429ae1f8ad81559641237a4f

## Revision 7 — 2026-08-05T23:52:36.688Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 123666
- hash: 68e42bacbdd58987d9a65392e884c627078c26a4d3e55ef47118a822f5d6cf94

## Revision 6 — 2026-08-05T23:47:10.346Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 118326
- hash: 33f6573a23216db254abfbed53fba9f642d485b2ae3ff9d2f9460a1dce1c46f9

## Revision 5 — 2026-08-05T23:45:17.827Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 116662
- hash: eaef4a25a68c6f33469bd56ee8b4ceb8a29593d88fb2ef8fcd8b625ef7a04a0c

## Revision 4 — 2026-08-05T23:36:32.719Z

- title: The obedience gap: why a model that reinterprets rules cannot hold terminal authority
- status: published
- bytes: 106069
- hash: 99d5397a1de48a9c7107428f1ed974d1cd019db4d5b3685b5b00a2573a45ebdc

## Revision 3 — 2026-08-05T23:06:04.528Z

- title: The obedience gap: what Anthropic claims and what its own research measures
- status: published
- bytes: 40404
- hash: 9b40f1c72a0cdfbf932e9c2bff76542079bf7a8b7ea0e784e3f7003cb3f66c60

## Revision 2 — 2026-08-05T23:02:37.937Z

- title: The obedience gap: what Anthropic claims and what its own research measures
- status: published
- bytes: 35948
- hash: 46bfb4f6bf1ddd7216ff57bd74e7d5f80f95c3fda99935ed03742cd254f4dc14

## Revision 1 — 2026-08-05T22:59:29.588Z

- title: The obedience gap: what Anthropic claims and what its own research measures
- status: published
- bytes: 33302
- hash: 857fab90551485de4e87b8f81c5c30715245fbbe642665562db775f67de5106e

## Revision 0 — 2026-08-05T22:57:50.187Z

- title: The obedience gap: what Anthropic claims and what its own research measures
- status: published
- bytes: 33302
- hash: ec1abcfa01ae8b8b4b11c46bf927ac16ff3b056f31768b9475ab7884f3ae036c