Question 1

Character.AI removed or restricted chat features you used? Is this new?

Accepted Answer

Character.AI has been relatively stable over the last 90 days. Character.AI has stayed stable, with minimal changes to its limits over time. Low recent constraint volatility — few meaningful restrictions in the last 12 months.

Question 2

Why is this happening — is it the tool or me?

Accepted Answer

Vanished features and suddenly guarded replies on Character.AI are a deliberate policy shift aimed at younger users, sitting on top of filtering that's been tightening for over a year. The decisive move came in late 2025: Character.AI said it would remove open-ended chat entirely for under-18 users, phasing down from a two-hour daily cap to zero effective November 25, 2025, and redirecting teens to creative-only features. That built on earlier changes — a modified, more conservative model for minors, a dedicated separate LLM for under-18s, and the removal of bot-response editing for everyone — so the platform now routes you to very different experiences by age. Your account's age status is the thing to check first, because it decides which model and which permissions you get; age assurance is deployed now, so a misclassified or teen-flagged account sees the restricted behavior no matter what its settings say. Open-ended roleplay that's simply gone rather than throttled, on an account that reads as under-18, is the policy, not a toggle you missed. An adult meeting heavier content blocks is running into the broader sensitive-content filtering instead — which one long-time user described as topics that worked fine six months ago now triggering instant blocks with unclear boundaries. Honesty about the remedies matters: verify and, if wrong, correct your age status through the account; accept that for younger users the removal of open-ended chat is permanent and not a setting; and if the constrained model no longer fits, this is the moment to weigh other tools. c.ai+ at $9.99 a month buys speed, memory, and skipping slow-mode, but it won't unlock content that policy has removed — and filtering is far likelier to keep evolving than to loosen.