Content Brain AI Content Quality Governance Framework

Document Type: Framework
Status: Active
Version: v1.1
Authority: Content Brain under HeadOffice governance
Applies To: All AI-Assisted Content Planning, Production, Editing, Review, Publishing Preparation, Refresh, Repurposing, Translation, Localisation And Programmatic Content Across MWMS
Parent: Content Brain
Enforcement Mode: Operational
Last Reviewed: 2026-06-15

Content Brain AI Content Quality Governance Framework

Purpose

The Content Brain AI Content Quality Governance Framework defines how MWMS may use artificial intelligence during content planning, production, editing, review and lifecycle management without sacrificing:

accuracy
usefulness
source integrity
audience relevance
trust
originality
editorial quality
compliance safety
human control

AI may support Content Brain work.

AI does not receive independent authority to:

invent facts
approve claims
make strategic decisions
approve offers
interpret compliance requirements finally
publish content
deploy advertising
send emails
post social content
operate autonomous production systems

This Framework applies to AI-assisted content across all Content Brain formats.

It is not limited to SEO content.

Core Principle

Content quality is determined by the usefulness, accuracy, integrity and suitability of the final asset.

It is not determined only by whether the content was written by:

a human
an AI system
a human using AI assistance
several contributors

AI-assisted content is acceptable only when the final asset is:

purposeful
accurate
sufficiently sourced
complete
audience-aligned
appropriate for its channel
edited
reviewed
approved
ready for its intended destination

AI involvement does not reduce the required quality standard.

AI involvement may increase the required verification standard where factual, legal, financial, medical, compliance, reputational or operational risk exists.

Framework Objective

The objective of this Framework is to ensure AI operates as a controlled production aid rather than an uncontrolled content generator.

The Framework must prevent:

fabricated facts
invented sources
hallucinated statistics
false authority
generic filler
near-duplicate content
unsupported claims
hidden assumptions
misleading summaries
incorrect product details
incorrect offer details
fake experience
fake testimonials
mass production without value
unreviewed publishing
content drift across related assets
automated content systems operating without authority

Scope

This Framework applies whenever AI assists with:

idea development
content planning
request clarification
brief creation
outline creation
structural blueprinting
drafting
rewriting
editing
summarisation
repurposing
translation
localisation
metadata
internal-linking suggestions
content refresh
content expansion
content correction
content merging
content-pack production
quality checking
publishing preparation
lifecycle review

It applies to:

Website Content

home pages
about pages
service pages
product pages
category pages
collection pages
landing pages
bridge pages
pre-sell pages
advertorials
reviews
comparisons
buyer guides
FAQ pages
local pages
authority pages
pillar pages
topic hubs
supporting pages
onboarding pages
knowledge-base pages

Article Content

educational articles
informational articles
commercial articles
affiliate reviews
comparisons
buyer guides
how-to articles
case studies
authority articles
research-led articles
search-led articles
refreshed articles

Social Content

Facebook posts
Instagram captions
Instagram carousels
LinkedIn posts
X posts
X threads
YouTube community posts
TikTok content
short-form scripts
engagement posts
social series
comment-response content

Scripted Content

YouTube scripts
YouTube Shorts scripts
TikTok scripts
Instagram Reel scripts
webinar scripts
video sales letter scripts
sales video scripts
explainer scripts
educational scripts
product demonstration scripts
podcast scripts
voiceover scripts
presentation scripts
training scripts
advertising scripts

Advertising Content

hooks
headlines
primary ad copy
descriptions
search ads
display ads
native ads
YouTube ads
Meta ads
TikTok ads
advertorial copy
pre-sell copy
creative briefs
CTA variations
message-match variations

Email And Newsletter Content

newsletters
promotional emails
educational emails
welcome sequences
nurture sequences
launch sequences
affiliate promotional sequences
re-engagement sequences
post-purchase emails
abandoned-cart emails
subject lines
preview text

Manuals Guides And Documentation

instruction manuals
user guides
implementation guides
playbooks
standard operating procedures
reports
white papers
ebooks
workbooks
checklists
lead magnets
onboarding packs
client documents
course lessons
training materials
internal documents
knowledge-base content

Content Packs

affiliate content packs
website packs
campaign packs
social packs
email packs
launch packs
product packs
documentation packs
repurposing packs

What This Framework Does Not Govern

This Framework does not independently govern:

broad research authority
offer approval
campaign approval
advertising budgets
bidding
targeting
capital decisions
statistical test validity
final compliance interpretation
final legal interpretation
final medical interpretation
final financial interpretation
publication authority
deployment authority
plugin implementation
worker activation
AI Employee activation
Brain Room routing
automatic M handoff

Those remain governed by the relevant Brain, human authority, framework, protocol or authorised system.

AI Assistance Levels

AI involvement should be visible and proportionate to the task.

Level 1: Administrative Assistance

AI may assist with:

formatting
organising notes
classifying information
converting approved material into templates
checking completeness
creating checklists
identifying missing fields
preparing structured records

This level carries limited creative or factual authority.

Level 2: Planning Assistance

AI may assist with:

brief preparation
outline creation
content architecture
asset planning
content-pack planning
section sequencing
question identification
objection identification
format adaptation

Planning assistance must use approved context and source material.

AI-generated planning does not authorise production automatically.

Level 3: Drafting Assistance

AI may assist with:

first drafts
section drafts
scripts
page copy
emails
social content
descriptions
summaries
calls to action
content variations

Drafting assistance requires editing and review before approval.

AI-produced drafts must never be treated as automatically accurate or complete.

Level 4: Editing Assistance

AI may assist with:

clarity
structure
grammar
readability
repetition reduction
tone alignment
formatting
headline refinement
CTA refinement
platform adaptation
consistency checking

AI editing must not silently alter:

facts
claims
prices
offer terms
evidence
legal meaning
compliance-sensitive wording
product instructions
technical requirements

Material changes must be reviewed.

Level 5: Review Assistance

AI may assist with identifying:

unsupported statements
possible inconsistencies
missing sections
repetition
tone drift
message mismatch
possible compliance risk
possible source gaps
cross-asset conflicts
incomplete publishing requirements

AI review is a supporting check.

It does not replace:

factual verification
specialist review
Compliance Brain review
legal review
technical review
Affiliate Brain review
Ads Brain review
human approval

Level 6: Lifecycle Assistance

AI may assist with:

identifying outdated sections
comparing versions
preparing refresh recommendations
identifying repurposing opportunities
identifying possible duplicate assets
identifying merge candidates
summarising performance signals
preparing lifecycle options

AI may recommend an action.

It must not autonomously:

alter live content
publish updates
merge live pages
retire pages
redirect URLs
change campaign assets
send communications

Acceptable AI Uses

AI may be used for:

ideation
planning
brief drafting
outlining
blueprinting
first-draft production
rewriting
editing
summarisation
repurposing
format adaptation
metadata drafting
question generation
FAQ drafting
content-pack coordination
content comparison
consistency checking
approved translation support
approved localisation support
refresh analysis
publishing preparation

AI use is acceptable only when it remains beneath:

approved purpose
approved sources
approved authority
required review
human control

Prohibited AI Uses

AI must not be used to:

invent facts
invent statistics
invent sources
invent quotations
invent studies
invent experts
invent endorsements
invent customer experiences
invent testimonials
invent product features
invent offer terms
invent prices
invent discounts
invent refund terms
invent rebill terms
invent scarcity
invent urgency
invent guarantees
invent search-performance results
fabricate firsthand experience
impersonate a real expert
create deceptive independent reviews
hide affiliate relationships
produce legal conclusions without authority
produce medical conclusions without authority
produce financial conclusions without authority
publish content without approval
deploy content autonomously
mass-produce content without demonstrated value
create near-duplicate pages from templates
convert assumptions into stated facts
conceal uncertainty
replace required research
replace required specialist review
replace human approval

Source-Control Principle

AI-assisted content must use an approved source base.

Possible sources include:

Research Brain outputs
Search Intelligence
Customer Brain intelligence
Strategy Brain direction
Affiliate Brain offer intelligence
Ads Brain requirements
Conversion Brain guidance
Creative Brain direction
Compliance Brain boundaries
Product Brain information
Ecommerce Brain information
business records
approved vendor material
verified product information
existing approved content
authorised human instructions
approved external sources

AI must not be asked to fill factual gaps by guessing.

Source Classification

Source material should be distinguishable where possible as:

confirmed fact
verified evidence
approved strategy
validated insight
emerging signal
working assumption
human instruction
creative direction
content interpretation
unverified material
prohibited source

The final content must not present:

assumptions as facts
interpretations as evidence
emerging signals as proven conclusions
creative language as factual proof
unverified material as confirmed information

Source Readiness Check

Before factual production, confirm:

Source Material Available:

Source Owner:

Source Date:

Source Status:

Source Gaps:

Facts Requiring Verification:

Claims Requiring Evidence:

Assumptions:

Interpretations:

Restricted Information:

Specialist Review Required:

If the source base is insufficient, the work should become:

Waiting For Research
Waiting For Evidence
Waiting For Product Information
Waiting For Offer Information
Waiting For Human Input
Waiting For Specialist Direction
Blocked

AI must not conceal a source gap by generating convincing unsupported copy.

Evidence-Control Rule

AI-assisted content must remain within the evidence provided.

AI must not:

strengthen a cautious source into a certain claim
generalise one example into a universal result
convert correlation into causation
remove material limitations
exaggerate effect size
misstate research conclusions
apply evidence to a population it did not cover
present vendor marketing as independent proof
create false consensus
create fabricated citations

Where evidence is uncertain, the content should preserve that uncertainty.

Hallucination-Control Rule

Every factual statement produced with AI assistance should be treated as potentially incorrect until confirmed by:

approved source material
verified business information
verified product information
verified offer information
approved research
specialist review
authorised human review

Higher-risk claims require stronger verification.

AI confidence is not evidence.

Fluent writing is not evidence.

Repeated AI agreement is not evidence.

Factual Verification Standard

Verification may include:

checking names
checking dates
checking prices
checking product details
checking offer conditions
checking refund terms
checking rebill terms
checking statistics
checking calculations
checking quotations
checking source meaning
checking legal or regulatory statements
checking technical steps
checking medical or financial statements
checking links
checking destination accuracy

Critical facts must be verified before approval.

Assumption And Interpretation Rule

Where content includes an assumption, inference or interpretation, it must not be disguised as a confirmed fact.

Possible handling includes:

label the assumption internally
qualify the language
request confirmation
remove the statement
seek specialist review
hold production

Examples of unsafe conversion include:

Working Assumption
→ stated as confirmed customer behaviour

Possible Explanation
→ stated as proven mechanism

Emerging Signal
→ stated as established market fact

Creative Angle
→ stated as product evidence

Audience And Intent Alignment

AI-assisted content must be produced for a defined audience and purpose.

Required checks:

audience identified
audience stage identified
intent understood
destination known
asset type appropriate
tone suitable
language suitable
CTA suitable
level of detail suitable
risk of misunderstanding considered

AI must not default to generic content written for everyone.

Usefulness Standard

AI-assisted content must perform a real job.

Possible jobs include:

answer a question
explain a subject
help complete a task
support a decision
compare options
reduce uncertainty
clarify an offer
guide a process
support a funnel
educate an audience
provide instructions
improve trust
prepare the audience for a next step

Content that adds no practical value should not be produced or published.

Completeness Standard

AI-assisted content must meet the completion standard of the approved request.

Content Brain must not provide:

an outline instead of a requested article
headings instead of a requested page
ideas instead of a requested social set
subject lines instead of a requested email sequence
bullet points instead of a requested manual
recommendations instead of a requested script
a brief instead of a requested finished asset
an asset list instead of a requested content pack

unless that is the approved output.

Originality And Information-Gain Standard

AI-assisted content should not merely paraphrase existing material.

Useful information gain may include:

clearer explanation
improved structure
stronger examples
better synthesis
updated information
practical steps
better comparisons
stronger decision support
transparent limitations
improved audience fit
better local or industry relevance
useful diagrams or tables
stronger internal relationships
correction of common misunderstandings

Originality does not require invented novelty.

It requires useful added value.

Duplicate And Near-Duplicate Control

AI must not be used to create:

many pages with only location names changed
many pages with only product names changed
many articles with the same structure and minimal variation
repeated affiliate pages with interchangeable copy
repeated social posts with superficial wording changes
repeated emails with no new purpose
multiple pages targeting the same intent without reason

Before scaled production, confirm:

each asset has a distinct purpose
each asset serves a distinct need
each asset has meaningful unique value
cannibalisation risk is controlled
internal relationships are clear
the production volume is justified

Voice Tone And Editorial Quality

AI-assisted content must align with the required:

voice
tone
audience
brand position
reading level
platform style
editorial standard

Review should remove common AI patterns such as:

vague introductions
empty enthusiasm
excessive summaries
repeated conclusions
unnecessary headings
artificial transitions
generic claims
repeated sentence structures
excessive qualifiers
predictable list construction
overuse of rhetorical questions
unsupported certainty
awkward formality
unnatural conversational language

AI language should not be retained merely because it is grammatically correct.

Editing Requirement

Every substantial AI-assisted draft requires editing before approval.

Editing should include:

structural editing
clarity editing
factual review
evidence alignment
audience alignment
intent alignment
tone alignment
voice alignment
repetition removal
filler removal
claim correction
trust improvement
CTA review
channel adaptation
formatting preparation
completeness checking

Editing must produce a stronger asset.

It must not become a cosmetic grammar check only.

High-Risk Content

High-risk content may include:

health
medical
disease
pain
mental health
sexual health
finance
income
investing
legal
safety
employment
environmental claims
regulated products
guarantees
scarcity
urgency
testimonials
before-and-after claims
sensitive personal decisions

High-risk AI-assisted content requires:

clear source control
evidence review
stricter factual verification
specialist review where required
compliance review where required
human approval
controlled publishing preparation

AI must not make final high-risk determinations independently.

Specialist-Review Triggers

Specialist review may be required when content includes:

medical claims
financial claims
legal statements
technical instructions
product-performance claims
safety guidance
regulatory statements
comparative superiority
guarantees
sensitive testimonials
significant reputational risk
uncertain evidence
disputed facts
material customer-impact risk

Possible reviewers include:

Compliance Brain
legal reviewer
medical reviewer
financial reviewer
technical reviewer
product reviewer
offer reviewer
brand reviewer
client
stakeholder

Human Oversight Rule

Human oversight remains mandatory under the current operating model.

Human review should confirm:

the requested asset was produced
the correct audience was served
the purpose was fulfilled
sources were used correctly
facts are accurate
claims are supportable
assumptions are not hidden
the asset is complete
the voice and tone are suitable
specialist review is complete where required
the correct version is approved
destination requirements are complete

Human review must not become a blind approval of fluent AI output.

Human Approval Rule

AI may not approve its own work.

Approval must identify:

approved asset
approved version
approving human
approval date where required
required changes
held sections
destination
publishing owner

Approved is not Publishing-Ready.

Publishing-Ready is not Published.

Delivery-Ready is not Delivered.

Website Content Controls

For AI-assisted website content, confirm:

page purpose
page position
audience
destination
existing-page check
duplicate-page risk
message consistency
internal links
CTA
metadata
page hierarchy
product or service accuracy
cross-page consistency
publishing preparation

AI must not create new website pages solely because keywords or topics are available.

Article Content Controls

For AI-assisted articles, confirm:

audience intent
content action
existing-content check
source base
information gain
evidence
structure
examples
internal links
external sources where required
CTA
metadata where relevant
refresh ownership

AI must not produce thin or repetitive articles merely to increase publishing volume.

Social Content Controls

For AI-assisted social content, confirm:

platform
audience
objective
hook
format
tone
length
CTA
visual requirement
disclosure requirement
claim safety
source relationship

AI must not copy the same generic post across every platform without adaptation.

Script Controls

For AI-assisted scripts, confirm:

script type
audience
platform
objective
duration
spoken clarity
pacing
scene requirements
shot requirements
on-screen text
evidence
claims
CTA
production notes

AI-generated scripts must be reviewed as spoken content, not only as written text.

Advertising Content Controls

For AI-assisted advertising content, confirm:

Ads Brain or authorised request
platform
audience
angle
hook
offer
destination
message match
claim boundaries
disclosure requirements
variation labels
test requirements
human approval

AI must not independently determine:

campaign strategy
budget
bidding
targeting
deployment
optimisation
scaling

Newsletter And Email Controls

For AI-assisted newsletters and email content, confirm:

audience or list
email type
sequence role
objective
subject line
preview text
body
CTA
offer details
links
disclosures
sending owner
follow-up logic
measurement

AI must not send newsletters or emails autonomously.

Manual Guide And Documentation Controls

For AI-assisted manuals, guides and documentation, confirm:

reader or user
required outcome
authoritative source
prerequisites
process order
technical accuracy
warnings
dependencies
examples
troubleshooting
completion criteria
update ownership
review date
delivery format

AI-generated technical instructions must not be accepted without appropriate factual or specialist review.

Content-Pack Controls

For AI-assisted content packs, confirm:

pack objective
audience
offer or project
required assets
shared sources
shared evidence
shared claims
shared disclosures
asset relationships
production order
consistent offer details
consistent messaging
cross-asset editing
specialist review
human approval
final versions
publishing sequence

AI must not create a pack by generating disconnected assets around the same topic.

Cross-Asset Consistency

AI-assisted multi-asset production requires a cross-asset review.

Check:

names
product details
offer details
prices
refund terms
rebill terms
mechanism
evidence
claims
disclosures
audience
terminology
CTA pathway
links
dates
version information

A correction affecting shared facts must be applied across every affected asset.

Refresh Controls

AI may assist with refreshing existing content.

Before refresh, confirm:

refresh trigger
current asset
useful content to preserve
outdated information
new sources
changed claims
changed product or offer details
internal-link changes
destination
approval owner

AI refresh must not:

rewrite useful content without reason
remove valid evidence
invent a performance problem
alter the page’s purpose without approval
change claims silently
publish changes automatically

A refresh is not automatically a formal experiment.

Experimentation Brain governs formal test validity where required.

Repurposing Controls

AI may assist with repurposing approved content.

Confirm:

approved source asset
new audience
new channel
new format
purpose
source meaning
evidence boundaries
claim boundaries
disclosure requirements
CTA adaptation
destination

Repurposing must not expand unsupported claims or present old material as new evidence.

Merge Controls

AI may assist with merging overlapping assets.

Confirm:

primary asset
secondary assets
material to preserve
duplicate material
conflicting material
final structure
internal-link changes
redirect requirements
retirement requirements
approval ownership

AI must not remove or combine live content autonomously.

Translation And Localisation

AI may assist with translation and localisation.

Translation requires review for:

factual accuracy
meaning preservation
terminology
cultural suitability
local spelling
local idiom
compliance differences
legal differences
units
currencies
dates
product availability
offer terms
CTA suitability

Pure machine translation without competent review may create:

factual drift
cultural mismatch
compliance risk
misleading terminology
broken trust

Translation changes language.

Localisation adapts the content to the destination audience.

They are not the same task.

Programmatic Content

AI may support controlled programmatic content where repeated structures are justified.

Programmatic production must define:

legitimate user need
distinct page purpose
unique source data
unique value
template controls
duplicate-content controls
quality threshold
review method
publication authority
monitoring
correction method
retirement method

Programmatic content is prohibited when it creates:

near-duplicate pages
pages with no distinct audience value
pages based on fabricated data
pages created only to capture keywords
low-value location substitutions
low-value product substitutions
mass content without review
automated publication without authority

Programmatic scale does not reduce the quality requirement.

Research Assistance Boundary

AI may help organise or summarise approved research.

It must not replace Research Brain when broad research is required.

AI research assistance must not:

fabricate sources
fabricate quotations
cite inaccessible material as confirmed
misrepresent study findings
treat summaries as primary evidence
hide uncertainty
convert unsupported internet claims into MWMS evidence

Research Brain remains the broad research authority.

Search-Specific Boundary

SERP analysis, search intent, entity planning and search-performance signals apply only where search is a material objective.

They must not be forced onto:

email
newsletters
social posts
internal documents
manuals
scripts
advertising
non-search website assets

Search Intelligence provides search evidence.

Content Brain produces search-led content.

AI must not invent current search conditions.

AI Review Record

For material AI-assisted assets, record where useful:

Asset Title:

Asset Type:

Purpose:

AI Assistance Level:

AI Used For:

Source Material:

Source Gaps:

Factual Verification Required:

Specialist Review Required:

Editing Owner:

Human Approval Owner:

Approved Version:

Destination:

Outstanding Risk:

Version And Provenance Control

AI-assisted content should preserve enough provenance to understand:

originating request
source material
production instruction
major assumptions
major revisions
specialist-review outcome
human approval
final version
destination

Provenance does not require saving every prompt permanently.

It requires sufficient traceability for material decisions and claims.

AI Output Status Model

Possible statuses include:

AI-Assisted Planning
AI-Assisted Draft
Draft Complete
Waiting For Sources
Waiting For Evidence
Waiting For Human Input
In Editing
Revision Required
Waiting For Specialist Review
Waiting For Human Approval
Approved
Publishing Preparation
Publishing-Ready
Delivery-Ready
Published
Delivered
Blocked
Held
Rejected
Cancelled

An AI-Assisted Draft must not be labelled Approved merely because it appears complete.

AI Content Quality Review

The quality review should ask:

Does the asset perform the approved job?
Is the audience clear?
Is the intent clear?
Is the asset complete?
Is the information accurate?
Are sources visible and sufficient?
Are claims supportable?
Are assumptions controlled?
Is the content useful?
Is the content specific?
Is the content original enough to add value?
Does it avoid generic filler?
Does it fit the channel?
Does it fit the voice and tone?
Has repetition been removed?
Has AI-style phrasing been edited?
Are disclosures present where required?
Is specialist review complete?
Has a human approved the correct version?

Minimum AI Content Quality Checklist

Use this for simple low-risk assets.

Confirm:

purpose clear
audience clear
destination clear
source material sufficient
facts checked
claims checked
no invented material
no hidden assumption
asset complete
useful
non-generic
voice and tone suitable
channel fit confirmed
editing complete
specialist review complete where required
human approval recorded
final version confirmed
publishing preparation complete

Stop Conditions

Stop production or publication when:

source material is insufficient
a required fact cannot be verified
a material claim lacks evidence
AI has invented a source
AI has invented experience
AI has invented product details
AI has invented offer terms
high-risk content lacks specialist review
audience or purpose is unclear
the asset is incomplete
the content is substantially generic
the content duplicates an existing asset without reason
cross-asset conflicts remain
human approval is missing
the final version is unclear
destination requirements are incomplete

Relationship To Content Brain Content Production System Framework

The Content Production System Framework defines how content moves through:

production
editing
specialist review
human approval
publishing preparation
lifecycle handoff

This Framework governs AI involvement within those stages.

Relationship To Content Brain Content Brief Template

The Content Brief Template defines:

Full Content Brief
Short Production Brief
Outline Or Blueprint
Direct Production Instruction

AI may assist with each level.

AI must not replace sufficient job definition.

Relationship To Content Brain Publishing Readiness Checklist

AI-assisted assets must pass the same Publishing Readiness controls as human-written assets.

AI involvement does not reduce:

evidence requirements
editing requirements
specialist-review requirements
human-approval requirements
destination requirements

Relationship To Content Brain Information Gain Framework

The Information Gain Framework defines how content should add meaningful value.

AI must not be used merely to rephrase existing content.

Relationship To Content Brain E E A T Content Trust Framework

The E E A T Content Trust Framework defines deeper experience, expertise, authority and trust controls.

AI must not fabricate any of these signals.

Relationship To Content Brain Editorial Consistency Framework

The Editorial Consistency Framework controls:

voice
tone
terminology
formatting
cross-asset consistency

AI-assisted content must comply with those controls.

Relationship To Content Brain Content Repurposing Framework

The Content Repurposing Framework governs how source assets are adapted across formats.

AI repurposing must preserve source meaning and evidence boundaries.

Relationship To Content Brain Content Optimization Framework

The Content Optimization Framework governs post-publication improvement.

AI may support analysis and drafting.

It does not receive autonomous authority to change live content.

Relationship To Content Brain Copy Map

This page is classified as:

Classification: Simplify Before Copy

Source Of Truth: MCR

Future Operational Use: AI Content Quality Review And Governance Module

Future Plugin Or UI: Possible AI-Assisted Content Quality Workspace

No plugin or UI implementation is authorised by this Framework.

Relationship To mwmscontentbrain.site

A simplified operational version may later support:

AI assistance level
source readiness
evidence status
assumption record
factual verification
quality review
specialist-review status
human approval
final version
publishing readiness

The operational version must not replace the MCR source.

Current Manual Operating Boundary

The AI-assisted content workflow remains human controlled.

Current restrictions:

no unrestricted AI content generator
no autonomous research worker
no autonomous drafting worker
no autonomous editing worker
no autonomous quality approval
no autonomous specialist approval
no autonomous compliance approval
no autonomous Affiliate Brain approval
no autonomous Ads Brain approval
no autonomous publishing
no autonomous WordPress changes
no autonomous social posting
no autonomous social scheduling
no autonomous newsletter sending
no autonomous email deployment
no autonomous advertising deployment
no autonomous page creation
no autonomous content refresh
no autonomous content merging
no autonomous page retirement
no active content queues
no Brain Room routing
no automatic M handoff
no AI Employee activation
no interference with M’s Research Brain work

The presence of this Framework does not authorise automation.

Future Operational Direction

A future AI-assisted content workspace may eventually support:

request context
source material
source classification
evidence requirements
AI assistance level
production instruction
draft status
assumption record
verification status
editing status
specialist review
human approval
version control
publishing preparation
lifecycle action
blockers
next steps

Future automation must remain beneath:

Content Brain Canon
HeadOffice authority
human approval
specialist controls
SIT oversight

Drift Protection

The system must prevent:

fluent AI output being mistaken for accurate output
AI confidence being treated as evidence
AI summaries being treated as primary research
assumptions being presented as facts
interpretations being presented as evidence
creative language being presented as proof
invented sources
invented statistics
invented studies
invented quotes
invented testimonials
invented customer experiences
invented product details
invented offer terms
fabricated expertise
fabricated firsthand experience
false authority
generic filler
thin content
duplicate content
near-duplicate programmatic pages
uncontrolled content scaling
keyword-driven content with no audience value
SERP requirements being forced onto non-search content
unreviewed translation
unlocalised translated content
AI drafts bypassing editing
editing being treated as approval
AI review replacing specialist review
AI review replacing human approval
approval being treated as Publishing-Ready
Publishing-Ready being treated as Published
repurposing expanding claims
refreshes altering live content without approval
AI merging or retiring live pages autonomously
AI publishing content
AI sending emails
AI posting social content
AI deploying advertising
inactive queues being treated as operational
operational pages becoming Canon
premature plugin or UI development
worker activation
AI Employee activation
Brain Room routing
automatic M handoff
interference with M’s Research Brain work

Architectural Intent

The Content Brain AI Content Quality Governance Framework exists to make AI a controlled quality and production amplifier.

It must help MWMS produce content that is:

faster to develop
easier to organise
easier to edit
more consistent
more complete
more reusable

without sacrificing:

truth
evidence
audience relevance
usefulness
originality
trust
specialist authority
human control

The Framework must support:

individual assets
websites
articles
social content
scripts
advertising
newsletters
email sequences
manuals
guides
documentation
content packs
refreshes
merges
repurposing
translation
localisation
controlled programmatic content

It must support current manual operation and later controlled implementation.

Future interfaces, workers and AI Employees must operate beneath this Framework.

They must not narrow, replace or bypass it.

Final Rule

AI may assist Content Brain.

AI does not become the authority.

AI-assisted content must not move into live use unless:

the purpose is clear
the audience is clear
the destination is clear
the source base is sufficient
factual statements are verified
evidence supports the claims
assumptions are controlled
the asset is complete
the content is useful
the content adds real value
voice and tone are suitable
editing is complete
specialist review is complete where required
human approval is recorded
the correct version is confirmed
publishing preparation is complete

AI must not invent missing information.

AI must not fabricate experience, expertise, authority or trust.

AI must not approve its own output.

AI must not publish its own output.

A fluent draft is not an accurate asset.

An accurate draft is not an approved asset.

Approved is not Publishing-Ready.

Publishing-Ready is not Published.

Delivery-Ready is not Delivered.

Humans retain final approval, publication, delivery and deployment control.

No premature automation.

No unrestricted AI generation.

No autonomous publishing.

No worker activation.

No AI Employee activation.

No Brain Room routing.

No automatic M handoff.

No interference with M’s Research Brain work.

Change Log

v1.1 — 2026-06-15

Replaced the original SEO-centred AI content framework with a complete multi-format AI content quality governance system.

Expanded the Framework to cover:

website content
articles
social content
scripts
advertising
newsletters
email sequences
manuals
guides
documentation
content packs
refreshes
repurposing
merging
translation
localisation
programmatic content

Added:

AI assistance levels
acceptable AI uses
prohibited AI uses
source controls
source classification
source-readiness checks
evidence controls
hallucination controls
factual verification
assumption and interpretation controls
audience and intent alignment
usefulness standard
completeness standard
originality and information gain
duplicate and near-duplicate controls
voice and tone controls
editing requirements
high-risk content controls
specialist-review triggers
human oversight
human approval
asset-specific controls
cross-asset consistency
research-assistance boundaries
search-specific boundaries
AI review records
version and provenance control
AI output statuses
quality-review questions
minimum quality checklist
stop conditions
current manual operating boundaries
future operational direction
expanded drift protection

Removed:

the broken content-reference artefacts
the universal SERP-analysis requirement
the assumption that every refresh is a formal SEO test
the incorrect universal routing of content tests to Ads Brain
the narrow focus on ranking risk as the primary AI-content concern

Clarified:

AI assistance does not equal autonomous production
AI confidence is not evidence
fluent writing is not factual proof
Research Brain remains the broad research authority
Search Intelligence applies only where search is relevant
Experimentation Brain controls formal test validity
AI review does not replace specialist review
AI cannot approve or publish its own work
human publication and deployment control remains mandatory

Aligned this Framework with:

Content Brain Canon v1.1
Content Brain Architecture v1.1
Content Brain Operating Model v1.2
Content Brain Workflow Map v1.2
Content Brain Content Production System Framework v1.1
Content Brain Content Brief Template v1.2
Content Brain Publishing Readiness Checklist v1.1
Content Brain Affiliate Funnel Support Map v1.2
Content Brain Affiliate Product Content Pack Framework v1.1
Content Brain Affiliate Product Content Pack Checklist v1.1
Content Brain SEO Content Brief Standard v2.1
Content Brain Page Registry v3.1
Content Brain Copy Map v2.6

v1.0 — 2026-04-26

Initial creation of the Content Brain AI Content Quality Governance Framework.

Defined the original AI-content controls covering:

helpfulness
accuracy
clarity
originality
trust
SEO risk
hallucination prevention
human oversight
information gain
SERP alignment
refreshes
programmatic content
international content
testing

Change Impact Declaration

Pages Created:

None

Pages Updated:

Content Brain AI Content Quality Governance Framework

Pages Deprecated:

None

Registries Requiring Update:

Content Brain Page Registry should show Content Brain AI Content Quality Governance Framework v1.1.

Content Brain Copy Map should show Content Brain AI Content Quality Governance Framework v1.1 where a version is recorded.

MWMS Architecture Registry update is not required unless it records individual Content Brain framework versions.

Canon Version Update Required:

Automation Status Change:

Plugin Or UI Status Change:

Queue Status Change:

No active queues authorised

Worker Status Change:

AI Employee Status Change:

M Handoff Required:

Research Brain Impact:

None

END CONTENT BRAIN AI CONTENT QUALITY GOVERNANCE FRAMEWORK v1.1