Content and Moderation Policy

DRAFT — NOT LEGALLY REVIEWED. This is the operational standard our moderators apply. It is also the document a court would most likely examine to judge whether we acted responsibly, so counsel should review it alongside the Terms.

Effective date: [EFFECTIVE DATE]

1. Principle

We publish allegations and patterns, clearly labelled as such — never findings of guilt. A report says "a traveller says this happened here", not "this business is criminal". Every design and moderation decision follows from that distinction.

2. What we publish

A report may be published where it:

• describes a method of deception or theft targeting travellers;
• is tied to a location (an area, landmark, transport route, or venue);
• describes something the reporter experienced or directly witnessed;
• is specific enough to be useful and, in principle, checkable.

3. What we do not publish

We reject reports that:

• name or identify a private individual — including by photograph, physical

description plus workplace, vehicle registration, or social media handle;

• allege serious crimes against a named person (assault, fraud) rather than describing a

method at a location;

• read as a competitor attack, a personal grievance, or part of a coordinated campaign;
• are pure price complaints — "expensive" is not deception; undisclosed or switched

charges are;

• contain slurs, discriminatory generalisations about nationality or ethnicity, or

harassment;

• rely on hearsay ("someone told me") with no first-hand basis;
• are unclear about the location, or so vague no traveller could act on them.

Nationality and ethnicity

Reports must describe conduct, never the ethnicity, nationality, or appearance of the people alleged to be involved. Descriptions like "watch out for [nationality] men near the station" are rejected outright. This is both an anti-discrimination rule and a legal risk rule.

4. Verification tiers

Every published report carries a visible label. The label is part of the publication — removing or obscuring it is not permitted.

Two things are shown together, because they answer different questions: where the entry came from (provenance) and what review it has had (status).

No entry is ever labelled as a verified finding of fact about a business.

5. Naming businesses — hard rule

Defamation requires a statement to be "of and concerning" an identifiable party. A report that names nobody is very difficult to sue over; a report that names a business is not. Staying at area and pattern level is therefore our primary legal protection, and it is enforced by the system rather than left to judgement.

The rule: reports are published at area, route, or pattern level by default. Naming a specific business is an exception that must be justified.

The API also screens every submission for language that reads like a business name and refuses to publish it silently — an entry that trips the screen must either be rewritten at area level or resubmitted with the flag, citations, and sign-off.

Other standing requirements:

• Describe the method, not motive. "Charged for items not ordered" is checkable;

"they are thieves" is not.

• Certain markets restrict naming entirely — see the per-market appendix, which

counsel must complete.

5a. Provenance must be stated accurately

Every report carries a label describing where it came from, and that label must be true:

Curated entries are never presented as traveller reports, and nothing is presented as a verified finding of fact. Misstating provenance to appear more authoritative — or to appear more like user-generated content than it is — is prohibited. It is both dishonest to users and, if it ever came under legal scrutiny, considerably worse than the underlying claim.

6. Automated tools

Automated scoring is used only to prioritise material for human attention. It never publishes, removes, or ranks content in front of users on its own. Every publication and removal decision is made by a person.

This is a deliberate commitment, not just current implementation: it keeps us outside the scope of rules on solely automated decision-making (GDPR Art. 22) and gives an accountable human decision behind every published allegation.

7. Moderator standards

Moderators:

• record a reason for every approve, reject, and dispute decision;
• do not moderate reports concerning a business they have any interest in;
• escalate anything involving a named individual, a legal threat, or a possible criminal

allegation against a specific person;

• work only within their assigned city scope, enforced by the system.

8. Corrections

If we publish something inaccurate, we correct it promptly and, where it materially affected a named business, note that a correction was made. We do not quietly delete and pretend it never happened.

9. Review

This policy is reviewed at least annually, and after any legal challenge, regulatory contact, or significant moderation failure.

Appendix — per-market restrictions

To be completed with counsel. Several countries in our current coverage have criminal defamation provisions or intermediary rules that may make naming a business materially riskier, and some may warrant area-only reporting or no launch at all. Countries currently in the database that need explicit advice include (non-exhaustive): Thailand, Turkey, UAE, Egypt, India, Vietnam, China, Indonesia, Morocco, Singapore.