Scatterlab said on 17 it has created a "Trust&Safety Team" reporting directly to the chief executive to strengthen user protection for its AI character chat service Zeta.
The new organisation will be led directly by Scatterlab CEO Jong-yoon Kim (김종윤) and will centrally manage content safety policy, AI model development and content review. It plans to assign half of its key AI developers to content review and to AI model development aimed at strengthening dialogue safety.
Scatterlab will upgrade its in-house AI model and content review system within the year to further break down usage ratings and strengthen safeguards by age. Its creative content service, Plot, will be reviewed across three stages - production, release and exposure - and it will tighten age verification to provide content tailored to users' ages.
It will also strengthen guardrails in its in-house AI model to prevent inappropriate utterances during conversations with AI. Zeta uses Spotwrite, a language model specialised in story generation that Scatterlab developed in-house. It will also introduce a function that periodically informs users that they are interacting with AI.
For teen users in South Korea, it plans to introduce within the year a function that limits daily usage time to 3 hours and sets a monthly basic payment cap at 100,000 won. It will also develop a function that provides appropriate guidance to users when sensitive content related to suicide or self-harm is detected.
Jong-yoon Kim said, "We will upgrade a protection system that spans policy, technology and operations and create an AI entertainment environment that people can use with peace of mind."