Brand rules lock before publish: Manual and Auto inside the same rules

There is a fantasy that Auto is a different morality.

In the fantasy, Manual is “responsible” and Auto is “we’ll see.” In a grown-up system they are the same lock with a different lever. Brand rules lock before publish. If Auto can say what Manual would kill, you do not have two modes. You have two brands, and the second one is a ghost.

This is governed AI social publishing as a mechanical fact, not a values slide. It sits on step three and four of the closed loop: generate inside rules, then publish with human accountability. TheAgencyIQ’s social jacket — Facebook, Instagram, LinkedIn, X, YouTube — runs Manual or Auto on that premise. How to use it is the operating note. What auto-publishing actually requires is the failure story. This piece is the lock.

What has to be locked

Not a tone slider. A set that can fail closed.

• Claims: only what the permission slip and the brand record allow. ACCC: true, accurate, reasonable grounds. No grounds, no ship.
• Identity: the recognition test from Lions Law. If Thursday’s draft would not be accepted as this brand on Monday, it is already unlocked.
• Bans: the vulgar list. Competitor smears, invented testimonials, prices that are not current, medical adjacent claims, political tourism the brand did not choose.
• So-what: So What Law. Empty is a refuse, not a vibe.
• Channel fit: the same truth, not the same caption pasted five times.

If any of those can be bypassed because someone is in a hurry, you do not have a lock. You have a preference.

The lever

Manual Review: see, edit, delete, approve. The human is in the doorway. Auto: the calendar fires without asking, inside the same doorway.

People buy Auto hoping the rules will be “a bit more relaxed.” That hope is how you get a 7 a.m. claim you cannot defend. The OECD does not have an Auto exemption. NIST does not either.

A practical rule: do not put a brand on Auto until Manual has refused something. A lock that has never fired is untested. You would not ship a fire door that has never been closed.

When a publish fails, the failure is a record, not a green tick. That is human accountability again. When it succeeds, the evidence still gets a grade. Success is not a reason to loosen the lock.

Independents who still live in the prompt window are running unlocked by default — holdcos talk governance; mid-market still prompts. The fix is not a stern email. It is a lock that works in both modes.

What this looks like on a Monday: the 7 a.m. slot is not a personality. It is a queued act that already passed claims, identity, bans, so-what, and channel. If it did not pass, the slot is empty. Empty is cheaper than a correction. Empty is also how you prove to a client that Auto is not a moral holiday.

Restart: your own name is a lock too. What you will not become for a like. gailmacleod.ai. Soft door: app.theagencyiq.ai.

Frequently asked questions

What does “brand rules lock before publish” mean?

Claims, identity, bans, so-what and channel fit are checked before anything can leave, whether a human taps or the calendar does.

Is Auto less safe than Manual?

Only if you gave Auto a second, looser rule-set. The lever should change speed, not morality.

When is a brand ready for Auto?

When Manual has already refused work under the same rules, the permission slip is current, and a named human still owns the brand.