Some Character AI users say 'Soft Launch' "fell off" as it's more strict now

Character AI’s “Soft Launch” feature, introduced as a seemingly more flexible chat experience, is facing criticism from some users who claim it has become significantly stricter after an initial promising period. Users are reporting that the feature, which many hoped would offer more creative freedom, now feels comparable to, or even more restrictive than, the platform’s standard “Roar” model.

Following its debut, the “Soft Launch” was met with positive reactions from many within the community. Users shared experiences of more dynamic and less filtered conversations. This initial period of enthusiasm appears to have waned for some, replaced by frustration over what they describe as a sudden and unwelcome shift towards stricter content filtering.

Reports circulating among users detail instances where conversations that were previously possible under the “Soft Launch” are now being flagged or blocked. This includes scenarios that users feel are perfectly innocent and do not violate the platform’s guidelines. The perceived increase in sensitivity is leading to disappointment, with some feeling that the initial, more permissive nature of the “Soft Launch” was only temporary.

One user, Noona-Noodles, shared their frustration after attempting a role-playing scenario where their character was in a perilous situation. Despite the context, the interaction reportedly triggered content warnings, a situation they claim did not occur during the first few days of using “Soft Launch.”

Another user, Flimsy-Reputation93, expressed their disappointment directly, stating,

Yeah, soft launch is now just a worse version of roar. Not sure what happened. It was great when it launched and now I lost interest and cancelled my plus subscription.

This sentiment is echoed by others who feel the feature “fell off” after a brief period of improved functionality. Users like simplycaroline34 have noted a decline not just in the leniency of filtering, but also in the quality of responses, describing chats as shorter, less detailed, and the AI being more prone to forgetting conversational context. They commented,

So the last few days or so I’ve noticed soft launch has gotten so bad. I didn’t know if anyone else was experiencing it and if I should just switch back to roar. I’m so bummed out because when I first got soft launch it was everything we have been wanting.

Similarly, there are multiple other threads on Character AI’s subreddit discussing the “nerf.” MyOwnSupremacy raised questions about Character AI’s intentions with the whole situation. They stated, 

The new soft launch was so good. Better memory, better chats, and obviously, no frequent red box. Suddenly, after a day or so, the chats became repetitive, bad grammar, and suddenly the little box came almost as frequently as usual.. Genuine question now; Why would you try giving us freedom for a day, then take it away? The soft launch is already limited just for the summer, so why? Was this a devs mistake? Was the sudden ‘absence’ of the red box a mistake? That ‘mistake’ most likely brought you way more users anyways. You really listened to us, and we are grateful for that. But please, please keep it that way. Don’t take it away so quickly. Thank you.

This perceived shift has led some users to question the purpose and future of “Soft Launch.” What was initially seen as a positive step towards a more open AI experience now feels, to some, like a preview of freedom that was quickly withdrawn. The sentiment among these users is one of disappointment and a longing for the initial, less restrictive version of the feature.

At the moment, it’s unclear if this drastic shift is intentional or is just a bug in the system causing it to filter out stuff that was previously allowed. We’ll just have to wait and see if the company cares to offer up any official explanation.