[[{“value”:”
As generative AI enters live-service games, conversational NPCs present exciting creative possibilities for those who choose to use them. However, it also presents a serious moderation challenge. In HoYoverse’s upcoming cozy life sim Petit Planet, players can visit a local cafe run by an AI-driven NPC called Nalo. Nalo will remember past conversations and use them to deliver dynamic, personalized responses.
However, during closed beta testing, some unsettling interactions involving heavy or self-destructive topics of conversation were noticed. Speaking to GameSpot at HoYoverse’s recent media week in Seoul, South Korea, developers Yinli Huang and Murphy addressed safety concerns, outlining how the studio is combining a series of moderation systems to ensure player safety.
When asked how the studio plans to prevent self-destructive prompts from slipping through, developer Yinli Huang emphasized that Nalo is not an open-ended chatbot, saying, “There are two sets of systems behind those neighbors. First of all, the NPCs that use the AI system, such as Nalo, will have a preset moral value and personality. As the relationship develops along with the players, it will develop a more customized interaction based on the input of each player.”
Because Nalo has such a strict moral compass, she will actively shut down toxic or manipulative prompts instead of blindly trying to appease the player. “Once players have said something inappropriate, Nalo will decline to respond to those requests or interact further,” Huang explained.
The developers recognize that the character’s refusal alone may not deter bad actors, nor does it support distressed players who may be using those systems to turn to someone when they need help. To address those problems, HoYoverse built a tiered escalation system into the game. “If that kind of interaction continues, we will put more limits [on] the player, such as a pop-up window notification and human intervention as needed,” Murphy explained. These measures will work alongside Petit Planet’s wider player-protection systems, which include account banning, in-game reporting, and restrictive measures to ensure children and vulnerable users remain safe.
Alongside automated systems and content filters, the development team understands the importance of granting players complete ownership over their privacy and involvement with the AI features. Instead of forcing the tech on everyone, Petit Planet’s AI features are entirely optional. “The players have full control of all the data the game generates, in terms of storage and deleting, and there’s a toggle in game that allows players to decide whether they want to use the AI features in the game.” However, HoYoverse couldn’t yet say what the experience would look like for players who choose to utilize that toggle, which is pretty concerning less than a month before the game launches.
What the team did say is that players who don’t want to use AI features will not be locked out of Nalo’s cafe entirely, nor will they be penalized in terms of progression or achievements. The cafe will remain as a social hub where players can gather with other players, hang out, and interact through the game’s standard, non-AI conversational features.
Pairing dynamic conversations with strict ethical boundaries, human monitoring, and the option to opt out of the feature mean that HoYoverse should be able to use the potential of generative AI without harming the pressure-free atmosphere that it hopes to achieve through a cozy life sim. As with many conversations around gen AI, it’s likely that the system will face scrutiny from some perspectives. Until the game launches and we can see the system in its finished form, it’s hard to say right now whether the AI features feel at home in the relaxing game.
“}]] [[{“value”:”As generative AI enters live-service games, conversational NPCs present exciting creative possibilities for those who choose to use them. However, it also presents a serious moderation challenge. In HoYoverse’s upcoming cozy life sim Petit Planet, players can visit a local cafe run by an AI-driven NPC called Nalo. Nalo will remember past conversations and use them to deliver dynamic, personalized responses.
However, during closed beta testing, some unsettling interactions involving heavy or self-destructive topics of conversation were noticed. Speaking to GameSpot at HoYoverse’s recent media week in Seoul, South Korea, developers Yinli Huang and Murphy addressed safety concerns, outlining how the studio is combining a series of moderation systems to ensure player safety.
When asked how the studio plans to prevent self-destructive prompts from slipping through, developer Yinli Huang emphasized that Nalo is not an open-ended chatbot, saying, “There are two sets of systems behind those neighbors. First of all, the NPCs that use the AI system, such as Nalo, will have a preset moral value and personality. As the relationship develops along with the players, it will develop a more customized interaction based on the input of each player.”
Because Nalo has such a strict moral compass, she will actively shut down toxic or manipulative prompts instead of blindly trying to appease the player. “Once players have said something inappropriate, Nalo will decline to respond to those requests or interact further,” Huang explained.
The developers recognize that the character’s refusal alone may not deter bad actors, nor does it support distressed players who may be using those systems to turn to someone when they need help. To address those problems, HoYoverse built a tiered escalation system into the game. “If that kind of interaction continues, we will put more limits [on] the player, such as a pop-up window notification and human intervention as needed,” Murphy explained. These measures will work alongside Petit Planet’s wider player-protection systems, which include account banning, in-game reporting, and restrictive measures to ensure children and vulnerable users remain safe.
Alongside automated systems and content filters, the development team understands the importance of granting players complete ownership over their privacy and involvement with the AI features. Instead of forcing the tech on everyone, Petit Planet’s AI features are entirely optional. “The players have full control of all the data the game generates, in terms of storage and deleting, and there’s a toggle in game that allows players to decide whether they want to use the AI features in the game.” However, HoYoverse couldn’t yet say what the experience would look like for players who choose to utilize that toggle, which is pretty concerning less than a month before the game launches.
What the team did say is that players who don’t want to use AI features will not be locked out of Nalo’s cafe entirely, nor will they be penalized in terms of progression or achievements. The cafe will remain as a social hub where players can gather with other players, hang out, and interact through the game’s standard, non-AI conversational features.
Pairing dynamic conversations with strict ethical boundaries, human monitoring, and the option to opt out of the feature mean that HoYoverse should be able to use the potential of generative AI without harming the pressure-free atmosphere that it hopes to achieve through a cozy life sim. As with many conversations around gen AI, it’s likely that the system will face scrutiny from some perspectives. Until the game launches and we can see the system in its finished form, it’s hard to say right now whether the AI features feel at home in the relaxing game.”}]]
Read MoreGameSpot – All Content
Discover more from GMDegens.io
Subscribe to get the latest posts sent to your email.


