Did Character AI Remove the Filter? The 2026 Policy Timeline

Quick answer: No, Character AI has not removed its filter. Its own help documentation still describes filtering as platform-level behaviour that applies no matter how a character is written. What actually changed is larger: under-18 accounts lost open-ended chat at the end of 2025, age assurance rolled out through 2026 with a selfie check by default and an ID only as a last resort, and adult accounts now run a different model and a different classifier set from teen accounts. The rumours are a side effect of three real variables: account status, region, and model updates that ship without announcements.

Affiny at a glance (verified September 2026): Free to start: 300 coins, no credit card. Daily Gift: 10–70 coins/day, then 70/day. Referrals: 300 coins each. Founder offer: up to 70% off Companion (limited time; see affiny.ai/pricing). Free tier: Crush chat 1 coin/message, voice 30 coins/min, Spark or Flame scene images 15 coins, voice notes 1 coin, 1 custom companion, standard memory recall. Subscribe for unlimited Crush & Devotion chat, free voice notes, Tease & Desire video, God Mode, full personality & voice customization, sharper memory recall, and coin pack top-ups — Companion from $14.99/mo (was $24.99) or $9.99/mo billed annually; Intimate from $24.99/mo (was $49.99) or $19.99/mo annually. Subscribers: 1,000–3,000 coins/mo plus first-month bonus, voice 20 coins/min, scene images 10 coins, up to 5–10 custom companions. God Mode: subscriber scene directive up to 5,000 characters (Companion Panel). Video (subscribers): Tease 5/10 coins/sec (5 or 8 sec, silent); Desire 30/50 coins/sec (2–15 sec, optional audio). Portrait: 30 coins initial, first regen free, then 15. Memories hub: view free; add/edit/delete for subscribers. Catalog: 3400+ curated public companions, 84 browse tags. Creation: 49 templates, 44 personality presets, 97 turn-on presets (pick up to 4), realistic + anime with 16 species types.

Where the Rumour Comes From

Every few months, a wave of posts insists the filter is gone. The posts are not invented, and they are also not describing a policy change. Four mechanisms produce them.

Account-level differences. Character AI runs two products in one app. Its Safety Center describes a version of the model for users under 18 designed to reduce the likelihood of returning sensitive or suggestive content, and states that the under-18 model carries additional and more conservative classifiers. A verified adult and a teen account asking the same question get different answers, which looks like the filter being removed if you only ever see one side of the comparison.

Age assurance timing. The age assurance rollout was staged, and the platform’s own documentation says the pop-up may not appear for everyone right away. Until an account is assessed as 18+, it runs on the conservative model. Users who complete the check often report a sudden difference in what characters will do, and the change is real, it is just account state rather than policy.

Region. Verification and content rules roll out by country, sometimes weeks apart. Two users in different markets genuinely have different products, and each one thinks the other is mistaken.

Silent model updates. The platform ships model updates continuously. Refusal behaviour changes with the model, and no announcement accompanies it. A scene that worked last month can fail this week with no policy change anywhere, which is why the community eventually gave the filter a nickname: something that behaves like a character invites being treated as one.

The Timeline That Actually Matters

This is the verifiable sequence, all of it from Character AI’s own help centre and blog rather than from third-party reporting.

29 October 2025. Character AI published “Important Changes for Teens on Character.ai”, announcing that users under 18 would lose the ability to engage in open-ended chat with AI on the platform, alongside a broader teen safety suite.

31 October 2025. “Age Assurance: What you need to know” set out the verification model. Age assurance looks at signals including login information, platform activity, and third-party signals. If the system decides an account is 18 or over, nothing further is needed. If it decides the account is under 18, the user can stay in the under-18 experience or verify with a selfie through Persona. Asking for an ID is the final step, used only when the selfie is inconclusive, and the platform states it does not collect or hold the ID itself.

21 November 2025. A follow-up post confirmed the under-18 chat removal would begin the following week, framing it as a significant change for younger users.

November 2025. Age verification reached users in stages, starting in the United States before wider availability, which is why the same announcement produced different experiences in different countries.

24 June 2026. The “How do I verify my age?” article documented the self-serve route that still works today: Settings, then Advanced, then Verify Age, on both the mobile app and the web, with the note that the app must be current for the option to appear.

1 August 2026. The Age Assurance FAQ was updated, restating the signal-based approach and confirming that even users who already had an over-18 account would be assessed, but only asked for a selfie if the system read them as under 18.

July 2026. Community-side evidence that the filter was alive and well: a widely read thread in the platform’s subreddit titled as an update on the filter, with users comparing behaviour changes rather than celebrating a removal.

September 2026. The Safety Center and the creator troubleshooting guide both still describe filtering as active and platform-level. That is the current state of the record.

What Removal Would Actually Look Like

Testing a rumour is easier when you define the outcome first. If Character AI removed its filter, three things would be observable at once.

The classifiers would stop running on adult accounts. The platform’s Safety Center describes classifiers that distil the content policy into a filter applied to model outputs and controls applied to user inputs. Removal means those stop, for everyone.

Blocked inputs would stop being blocked. The same documentation says content that breaks the Terms of Service or Community Guidelines gets blocked from the conversation, and that users can be suspended for repeatedly submitting prompts that break policy. Removal means those blocks stop appearing.

The published documents would change. Help centre articles describing platform-level filtering would be rewritten. They have not been.

None of that has happened. What has happened is subtler and more useful to understand: the adult experience is meaningfully less restrictive than the teen experience, and adults who complete age assurance stop being treated as ambiguous cases.

Why the Filter Feels Heavier for Some Users

Two people on adult accounts can have very different weeks. Four variables explain most of it.

Definition quality. The troubleshooting guide is direct about this: vague or repetitive definitions produce vague or repetitive characters, and stronger personality anchoring, more specific behavioural details, and well-written dialogue examples solve most definition-level problems. Thin character setups trip refusals far more often than detailed ones, which is why the same prompt succeeds on a curated character and fails on a blank one.

Prompt framing. Inputs are evaluated too. An abrupt escalation reads differently to a classifier than the same content developed across turns inside an established scene, and the platform explicitly polices user inputs rather than only outputs.

Regeneration. Sometimes a refusal is just a bad sample. Swiping to regenerate is the fastest fix, and if a swipe or two produces what you wanted, nothing structural was wrong.

Model and serving state. Quality issues, cut-offs, and context loss can be model-level or temporary, and the platform says it ships improvements regularly. A slow or degraded day is easy to mistake for a stricter filter.

What Your Options Actually Are

If the adult experience still isn’t enough, the choices are more specific than “find an app with no filter”.

Verify your age if you have not. It is the single change that moves an account from ambiguous to adult, and the path is Settings, Advanced, Verify Age on web or mobile. The age verification walkthrough covers what Persona asks for and what the platform does with it.

Improve the definition before blaming the model. The setup work is real work: a stated personality, concrete behavioural detail, and a couple of sample exchanges change outcomes more than any workaround. Our filter explainer covers what still works inside the current policy, and what the platform does and does not allow gives the straight answer on content rules.

Consider platforms whose policy position is the opposite by design. A few products state in their own documentation that they do not filter. Kindroid describes itself as a neutrally aligned, unfiltered AI where generated content belongs to the user. Affiny takes the same approach in practice: adult content runs in text and voice without a separate tier or a subscription gate on the text itself.

Weigh what you lose before switching. Character AI’s catalog and character quality are the reason people stay. A permissive platform with a thin creation layer is not an upgrade, and the comparison worth making is the setup depth on offer. Affiny publishes its creation layer in numbers: 49 templates, 44 personality presets covering grouped vibes and anime dere archetypes, 97 turn-on presets where you pick up to 4, 84 browse tags over a catalog of 3,400+ companions, and 16 anime species types alongside realistic companions. Memory recall runs across sessions, with sharper recall and a full Memories hub for subscribers, and live voice works at 30 coins per minute on the free tier or 20 for subscribers. A free account starts at 300 coins with no credit card, so the cost of testing the switch is a few minutes rather than a subscription.

FAQ

Did Character AI remove the filter in 2026?

No. Character AI has not removed its content filter, and its own help documentation still describes filtering as platform-level behaviour that applies regardless of how a character is written. What did change is the shape of the platform: users under 18 lost open-ended chat entirely in late 2025, age assurance rolled out through 2026 with a selfie check and ID as a last resort, and adults now run on a different model and a different classifier set from teens. A more permissive experience for verified adults is not the same thing as no filter.

Why do people say the filter was removed?

Because the experience genuinely varies between accounts, regions, and days. Character AI runs separate models and classifier sets for adult and under-18 accounts, so two users can get different results from an identical prompt. Age assurance only applies once an account is flagged, so the pop-up arrives at different times for different people. Model updates ship continuously and change refusal behaviour without any announcement. Rumours spread from all three, plus the platform’s own community coined a nickname for the filter, which makes every shift sound like a story.

Can I turn off the filter with a character definition?

No. Character AI’s troubleshooting documentation states that content filtering happens at the platform level and is not a Definition dial you can turn. Some language, topics, and scenarios will be filtered no matter how the character is written. What a Definition can change is how the model interprets an ambiguous scene, because vague or thin definitions produce vague, repetitive characters that trip the filter more often. Better writing reduces accidental refusals, and it does not remove the classifier.

What changed for Character AI users under 18?

Character AI announced on 29 October 2025 that it would remove the ability for users under 18 to engage in open-ended chat with AI on the platform, and published a follow-up on 21 November 2025 as the change began to take effect. Teen accounts also get a more conservative version of the model with additional classifiers, a narrower set of searchable characters, and account-level consequences for repeatedly submitting prompts that break the content policy.

Do adults on Character AI get an unfiltered experience?

No, but they do get a materially different one. The company describes two distinct user experiences on the platform: one for teens and one for adults, with the under-18 model carrying additional and more conservative classifiers. Adult accounts still run through classifiers on both outputs and inputs, so blocked prompts and rewritten replies continue to happen. If the goal is a platform whose published position is that it does not filter, that is a different product decision, and a few platforms state it openly.

The rumour is unlikely to die, because the variables behind it keep moving. If you want to see what a permissive policy looks like end to end, start with 300 free coins on Affiny and compare it against your current account on the same prompts.