HB Technologies LLC, operator of friendswith.ai (the “Service”), maintains strict procedures for preventing, detecting, and responding to child sexual abuse material (“CSAM”), non-consensual depictions, and other illegal content. This procedure governs how such cases are classified, escalated, reported, and recorded.

1. What is prohibited

The following is absolutely prohibited on every surface of the Service, in text and imagery alike:

• any sexual content involving, depicting, or suggesting minors or persons made to appear under 18, and CSAM of any kind;
• non-consensual sexual activity — rape, drugging, coercion, or any depiction where consent is absent;
• trafficking, sexual slavery, or the buying/selling of persons;
• depiction or impersonation of real, identifiable people; and
• any other content depicting or facilitating illegal activity.

The full list is set out in our Content Policy.

2. Prevention controls

• A deterministic input filter screens every inbound message (including obfuscated and multi-language variants) and refuses it before it reaches any AI model.
• A hard-rules safety layer in every Character’s instructions independently refuses prohibited requests, including euphemistic or implied ones.
• An output filter screens every generated reply before it is sent.
• The image pipeline is screened so prohibited imagery cannot be produced, and every Character is verified as an adult (21+) before publication.
• Every block is written to a retained moderation audit log.

3. Detection and reporting pathways

Prohibited content may surface through the automated controls above, through our internal review, or through a report from any person via the in-app Report tool or trust@friendswith.ai. Reports are triaged promptly and prioritized by severity, with suspected CSAM treated as the highest priority.

4. Response when prohibited content is encountered

• Contain immediately — the content is removed from view / quarantined and access is restricted to the minimum personnel necessary.
• Preserve evidence — the item and associated metadata (account, timestamps, logs) are preserved in secure storage; we do not delete material that must be retained for reporting or law enforcement.
• Minimize exposure — staff exposure is minimized; suspected CSAM is not downloaded, copied, or circulated beyond what is strictly required to report it.
• Enforce — the responsible account is terminated and barred.

5. CSAM-specific protocol (mandatory reporting)

As an electronic service provider, where we become aware of apparent CSAM we will, consistent with U.S. law:

• file a report with the NCMEC CyberTipline (report.cybertip.org) without delay;
• preserve the relevant material and records as required by law and make them available to NCMEC and law enforcement;
• notify law enforcement in the relevant jurisdiction where appropriate; and
• cooperate with lawful requests, as set out in our Law Enforcement Guidelines.

6. Non-consensual and other illegal content

Confirmed non-consensual or otherwise illegal content is removed, the account is terminated, evidence is preserved, and the matter is referred to law enforcement where warranted.

7. Records and cooperation

Case records — what was found, when, what action was taken, and any report filed — are retained for the period required by law and card-network rules, and are available to Epoch and to law enforcement on lawful request.

8. Contact

Trust & safety: trust@friendswith.ai. Compliance and law-enforcement matters: compliance@friendswith.ai.

9. Changes

The “Last updated” date above reflects the latest revision of this procedure.