Trust & safety

All moderation runs through PaxMod.

We don't staff a content team to sit in the middle of your friendships. Where friendAI needs trust and safety — public companions, shared chats, reports — we hand it to PaxMod. They're based in Adelaide, backed by the Australian Institute for Machine Learning, and this is the whole company: the infrastructure for online safety, not a side feature we bolted on.

“We build the infrastructure for online safety.”

Real-time API

Every public message is scored before it lands, with low-latency checks built for live chat.

Text and images

Toxicity, hate, harassment, CSAM, self-harm, spam, and PII — plus the categories we actually need.

Rules, then people

Custom thresholds for our community, and a human review queue when the edge cases matter.

Built to be trusted

SOC 2 Type II, GDPR-ready, encrypted in transit. They don't keep message content longer than processing needs.