One real, disclosed identity in the Comment Engagement Desk
Satisfies:
- UCPD Annex I point 22
- DMCC Act 2024 Part 4
- FTC 16 CFR Part 465
- FTC Act section 5
Trust Center / White-Hat Standard
Trust Center
The white-hat
standard.
GEO can be done two ways, and only one of them survives contact with a better model.
One improves how accurately and completely a brand is represented, using verifiable information published in places models legitimately read. The other exploits model weaknesses, fabricates consensus, or hides intent. The first gets better over time as models improve. The second gets caught. Hordus does the first, exclusively, and structures its products so the second is not available even if a client asked for it.
What we do
Six practices.
Publish verifiable information
Every factual claim we publish on a client's behalf traces to a source the client can stand behind.
Improve structure, not just presence
Schema, entity markup and clear hierarchies, helping models parse what is already true.
Correct the record at the source
Where AI misdescribes a client, we fix the underlying sources rather than trying to trick the model.
Disclose affiliation
Brand-affiliated content says so, prominently and in the structured data.
Human review before publication
Every asset, without exception, approved by a named person.
Respect platform terms
Site terms and platform policies are a ceiling, not an obstacle to route around.
What we never do
Ten hard stops.
✕
No fabricated or multiple identities, no persona libraries, no identity rotation✕
No simulated grassroots support or manufactured consensus✕
No engagement gaming: no vote manipulation, like-boosting or coordinated amplification✕
No auto-generated conversations between accounts✕
No cloaking: crawlers never receive substantively different content from humans✕
No prompt injection, jailbreaking or exploitation of model vulnerabilities✕
No link schemes, private blog networks or content farms✕
No CAPTCHA solving, login bypass or terms-of-service evasion✕
No undisclosed political or commercial messaging✕
No unreviewed AI output published to any audienceOut of scope, not deprioritized
A design-review stop, not a backlog item.
If a requested capability would require any of the above, it is out of scope. Not deprioritized, not under review. Our product specifications treat such requests as a design-review stop rather than a backlog item.
Compliance mapping
Which instrument each practice satisfies.
Human approval of every comment, with audit log
Satisfies:
- 16 CFR Part 465: no fabricated or insider-originated content presented as independent
Site terms as a ceiling
Satisfies:
- Platform terms of service
- Contract law generally
Content parity between renderings
Satisfies:
- Search and AI platform cloaking policies
- FTC Act section 5
Published methodology for every score
Satisfies:
- The substantiation principle under FTC Act section 5 and equivalents