Content Governance & Moderation

Content Governance & Moderation

1. What it is

Content Governance & Moderation are two admin surfaces that keep your tenant’s content safe and current. Content Moderation screens what people post — comments, chat messages, AI prompts — as they post it, and blocks, holds or flags it. Content Governance checks content that is already published and files a finding (a flagged issue) for each problem: overdue for review, a broken link, expired but still live, no owner. Each finding goes to the person or team that owns the content.

• Enablement: both live under Admin → General Settings, with no separate licence. Governance scanning is switched on per app on its Settings tab.
• What it is not: moderation does not judge old content, and governance does not screen new posts. Moderation asks “should this be published?”; governance asks “is what we published still right, and who answers for it?”

2. Standing it up

Content Moderation

1. Go to Admin → General Settings → Content Moderation, then open Settings.
2. Under What to Moderate, pick one option. The default is Moderate only flagged content: posts publish normally and users can report them. The others are Moderate all content (every new post and comment is held for approval) and Keyword + AI moderation (content is flagged when it matches a keyword or the AI detects an issue).
3. For keyword screening, add words to Custom Keywords, separated by commas. A built-in library of offensive words is checked as well.
4. To add AI screening, tick Enable AI content moderation. It starts in Observation mode, which records what the AI would flag without blocking anything. Switch to Live when you trust the results.
5. Under Which Users / Teams?, the default is All users. You can pick Only specific users or Exempt specific users / teams instead. Admins are exempt from screening on most surfaces.
6. Check the Queue tab. Held and reported items appear there for review.

Content Governance

1. Go to Admin → General Settings → Content Governance, then open Settings. Each governed app has its own card.
2. Tick Enable automated governance scanning for each app you want checked. It starts on for Ask AI, Drive, Forms, Policy Hub, Training and Workspace, and off for the rest. Sites, Safety Hub, Inspections and Contracts are switched on from their own settings; their card links there.
3. Set Review Interval (days) where the card offers it. Some apps set their own interval, and the card shows it read-only.
4. Name owners. Open a policy, SOP, wiki page, library item, form, course, handbook, knowledge-base article or site page and fill in its Owner panel. Owned content does nothing extra until governance is on for its app — without the scan, nobody is reminded.
5. Wait for the weekly scan (every Monday at 07:00 UTC), or click Run Scan Now on the Findings tab. You can’t start another within 5 minutes of the last run.
6. Check the Findings tab for results.

3. How it fits together

Owner, owning team and author. Every governed item can name who answers for it. The Owner panel on the item’s edit or settings screen has a person field and an Owning team field (one of your groups). You can fill in either or both. If both are empty, the item’s author answers for it until an owner is named. Only someone who can open the item can own it; for a team, everyone in it must be able to open it.

The reviewer chain. When a scan files a finding on owned content, the reminder goes to the first of these that applies:

1. the named owner, while they are an active member;
2. otherwise, the owning team’s members (up to 10 of them);
3. otherwise, the author, while they are an active member;
4. otherwise, the app’s own reviewers — by default your admins (up to 5).

Only people who can open the item are told. A worked example: a Policy Hub policy names no owner but has Owning team set to “HR Operations”. When its review falls due, the members of HR Operations who can open that policy are reminded. If everyone in the team has left, the reminder goes to the policy’s author, and failing that to your admins.

Content I own. Every person has a page at avatar menu → Content I own. It lists what they own in each app, whether they own it as Owner, as a Member of a team, or as Author (no owner named). It shows each item’s review date and any open findings. It also lists findings assigned to them, Possible duplicates, and anything Unpublished automatically. Items they own but cannot open are counted, never named, with a prompt to ask an admin.

Governed apps. Each app decides what it checks. The apps with governance today are AI Front Office, Ask AI, Communications Hub, Contracts, Drive, Forms, HR Files, Inspections, Libraries, Mango GTM, Mango IQ, Onboarding Hub, Policy Hub (policies and handbooks), Recorder, RFP Desk, Safety Hub, Service Desk, Sites, SOP Hub, Training, Wikis, Workforce Bots and Workspace.

Findings come in types. On Content I own, common ones read No owner, Owner has left, Owner can’t open it, Due for review, Review overdue, Expired, still live, Broken link and Possible duplicate. The admin Findings list names a few differently — Inactive owner, Stale content, Expired but live. The duplicate check runs weekly across policies, SOPs, wikis, site pages and library items.

Unanswered questions (Ask AI). Ask AI groups questions it could not answer from company knowledge. A group is shown only once at least 5 different people asked it. Each open group becomes an Unanswered question finding. The scan assigns it to a topic expert for the question’s theme who can open it; with no expert, it stays unassigned and your admins are reminded. The owner writes the answer with Create an answer, if they are allowed to write Verified Knowledge. An owner who isn’t an admin opens a read-only page for that one group from their reminder or from Content I own; admins work every group at Ask AI → Agent Guidelines → Questions. An answer from a non-admin waits for admin approval in Verified Knowledge, and no reminder goes out meanwhile. Approval closes the group and resolves its finding; a rejection hands the question back to its owner. Every Monday, before the scan, Ask AI re-asks each open group and closes any that company knowledge now answers.

Moderation modes. Moderate only flagged content acts only on user reports. Moderate all content holds every post for review. Keyword + AI moderation holds or blocks only what triggers a rule. AI Observation mode logs without acting; Live mode blocks or holds.

Severity escalation. Open Low findings become Medium 30 days after they were found; Medium findings become Critical 60 days after they were found. Both are adjustable per app. A finding climbs at most one level per weekly scan. High findings, snoozed findings and possible duplicates never escalate.

Boundary with offboarding and groups. When an Offboarding plan removes someone, the policies, handbooks, SOPs, wikis, knowledge-base articles, library items, forms and courses they owned — or wrote with no owner named — are handed to their manager, else an admin. Only items the new owner can open move; the rest stay and are flagged Owner has left. Site pages are handed on by Sites itself. A group that owns content can’t be deleted until that content has a new owner.

4. Running it

Working the governance findings list

1. Go to Admin → General Settings → Content Governance → Findings.
2. Filter by app, issue type, severity or status. Snoozed findings are hidden until you tick Show snoozed.
3. For each finding, choose Acknowledge, Resolve, Dismiss, Assign (hand it to a person) or Snooze (hide it until a date; it stays open).
4. Select several findings for Bulk Acknowledge, Bulk Resolve or Bulk Dismiss.
5. Click Export CSV to download the filtered list.

Handing someone’s content to someone else

Use this when a person changes role, or left without an Offboarding plan.

1. Go to Content Governance → Settings and scroll to Reassign content ownership. People who have left and still own content are listed there.
2. Pick Reassign from (anyone, including people who have left) and Reassign to (an active member).
3. Click Reassign all. Everything the new owner can open moves to them, and their open ownership findings are settled. You’re told how many items stayed behind. The new owner gets a notification pointing to Content I own.

Site pages are reassigned from each site’s settings instead.

Reviewing the moderation queue

1. Go to Admin → General Settings → Content Moderation → Queue.
2. Filter by source, issue type or severity.
3. Approve publishes the item; Reject removes it. The author is told either way, and you can add notes.
4. Below the queue, unscreened publishes from the last 7 days are listed. These went live because the automated screen did not answer in time. They are a record, not a to-do.

Reading the analytics

• Content Governance → Analytics shows findings over time, by issue type, app and severity, and time to resolve. Its Gap Analysis view shows the searches that found nothing, for apps where gap analysis is on.
• Content Moderation → Analytics shows flagged content by source and severity, top flagged users, moderator activity, review time and AI security activity.

5. Settings

Content Moderation — Admin → General Settings → Content Moderation → Settings

Setting	Default	What it changes
What to Moderate	Moderate only flagged content	Whether content is screened on report, always held, or screened by keywords and AI
Custom Keywords	Empty	Extra words to catch, added to the built-in list
Enable AI content moderation	Off	Whether the AI screens content
AI Mode	Observation	Whether AI detections are only recorded, or block and hold content
Harm Categories: Harassment & bullying	On	Whether the AI screens for harassment
Harm Categories: Discrimination	On	Whether the AI screens for discrimination
Enable negativity detection	Off	Whether the AI scores negativity
Negativity Tolerance	50	How sensitive negativity scoring is
Which Users / Teams?	All users	Who is screened
Notify moderators when new items enter the queue	On	In-app and email alerts to moderators
Notify content authors when their content is approved or rejected	On	Email to the author on a decision
Hide personal information in AI answers	On	Hides Social Security, card and bank account numbers in AI answers. Emails and phone numbers are hidden when an answer holds three or more of them

Content Governance — Admin → General Settings → Content Governance → Settings (per app)

Setting	Default	What it changes
Enable automated governance scanning	On for Ask AI, Drive, Forms, Policy Hub, Training, Workspace; off for the others	Whether the weekly scan checks this app
Review Interval (days)	90 for Libraries; set by the app elsewhere	How long content may go unchanged before it is due for review
Expire content after (days)	Blank (never)	Age at which content counts as expired, where the app supports it
When content expires	Flag it for review	Flag only, or also unpublish it 14 days after it was flagged if still unresolved
Check broken links	On, where the app offers it	Whether external links are tested
Notify on governance review	On	Whether owners, teams and reviewers get reminders
Enable gap analysis	Off	Whether search logs are analysed for missing topics
Notify on gap summary	On	Whether admins get the gap-analysis digest
Low → Medium (days)	30	When open low findings escalate
Medium → Critical (days)	60	When open medium findings escalate

Each person can also mute the Ask AI reminder in their notification preferences, under Ask AI unanswered questions. It is on by default, in-app and by email. The same switch mutes the notice that an answer they wrote wasn’t approved.

6. More help

• Content Governance FAQ — specific questions about scans, owners, findings and moderation decisions
• Ask AI Overview — how Ask AI answers from Verified Knowledge
• Audit Logs & Activity History — who changed what, including moderation decisions
• Ask AI — the assistant answers questions about Content Governance & Moderation from these articles.