Is Character AI Shutting Down? The Complete Guide to What's Actually Happening

Is Character AI Shutting Down?

No, Character AI is not shutting down. The platform is fully operational as of April 2026, and the evidence points to a company that is very much alive — even if it doesn't feel the same way it used to.

The numbers tell a clear story. Character AI pulled 194.62 million monthly website visits in March 2026, up 31.67% from February, with an average session duration of nearly 20 minutes.

Revenue hit $32.2 million in 2025 — more than double the $15.2 million from 2023 — and projections put it at roughly $60 million for 2026. Google invested $2.7 billion in the company in August 2024, hiring co-founders Noam Shazeer and Daniel De Freitas in the process. Character AI's own roadmap reports an 85% reduction in outages from the prior year.

Is Character AI going to shut down? Not with those numbers. Is character ai being shut down or closing down by some external force? Also no.

The platform is not being forced offline — it's being forced to change, which is a very different thing. Did character ai shut down at any point? It did not. If you just hit a load error and want the troubleshooting rundown, our guide on why is character ai not working covers the common temporary issues.

But here's where the nuance matters. Character AI's monthly active users declined from 28 million at the mid-2024 peak to approximately 20 million by early 2025 — a loss of roughly 29% of peak traffic.

The platform is alive, but it is losing people. The real question is not whether Character AI is dying — it's whether it's becoming something you no longer recognize.

What Does an Actual AI Platform Shutdown Look Like?

When AI companion platforms actually shut down, they announce it. Butterflies AI shut down on February 28, 2026, despite raising a $4.8 million seed round just 20 months earlier — citing changes in external services that made operations impossible.

Moemate shut down in early 2025 due to unsustainable GPU costs, deleting all user data after peaking at over 6 million people. Both followed the same pattern: official announcement, data deletion, service termination.

Character AI has none of these signals. Not one.

What Specific Changes Has Character AI Made Recently?

Character AI has made more than a dozen significant changes since mid-2024 — and the pace has only accelerated. If you've been wondering why did character ai change or whether character ai has gotten stricter, the timeline below shows exactly what happened and when. For more on filter-specific questions, our guide on does character ai allow nsfw breaks down every filter change in detail.

The Full Timeline

• August 2024 — Google invests $2.7B and hires co-founders Shazeer and De Freitas; leadership vacuum shifts platform direction.
• September 2024 — Parents of teens who died by suicide testify before Congress; public pressure intensifies.
• October 2024 — FTC launches formal inquiry; federal-level investigation into platform safety practices begins.
• October 2024 — “October 15 shutdown” rumor spreads on social media with no factual basis (see FAQ).
• Late September 2025 — Disney sends cease-and-desist regarding IP characters, signaling the IP compliance crackdown to come.
• October 13, 2025 — California SB 243 signed into law — first U.S. state law mandating AI companion safety safeguards.
• October 29, 2025 — Character AI announces end of open-ended chat for under-18 users.
• November 24–25, 2025 — Phased teen chat restrictions begin; 2-hour daily cap imposed.
• November 25, 2025 — Stories feature launches as teen replacement for open chat.
• December 2025 — Chat history expanded 10x (100 to 1,000 messages); Charms system rolls out on mobile.
• January 2026 — Character AI and Google settle five teen-related lawsuits; financial terms undisclosed.
• February 18, 2026 — “Moderatedpocalypse” mass automated bot removal sweep; thousands of community-built characters deleted overnight.
• April 2026 — Mandatory face-scan age verification (Persona) introduced for all users, not just teens.
• April 14, 2026 — PipSqueak 2 model released with memory redesign claims.
• May 2025 (ongoing) — “Soft Launch” behavioral dampening layer introduced; hidden emotional resets every 8–10 messages.
• Mid-2025 — AvatarFX, Scenes, and Streams launch, signaling a strategic pivot from pure chat to multimedia.

Did Character AI update? Constantly. Has character ai changed? Beyond recognition for many people. And the changes have not been limited to one year — is character ai shutting down in 2025 was already a common search, because this has been building for over 18 months.

What Changed for Adults vs. Teens

The teen chat restrictions that rolled out in November 2025 eliminated open-ended conversation for people under 18. But the ripple effects hit everyone.

Filter tightening, the Soft Launch behavioral dampening, and the Charms monetization system all affect adult accounts too. Has character ai gotten stricter? Yes — and it's not just a teen safety measure. Filters now trigger on emotional roleplay that has nothing to do with explicit content, which means the experience has changed for everyone regardless of age.

The “Moderatedpocalypse” and What It Erased

On February 18, 2026, Character AI ran a mass automated copyright and IP compliance sweep that removed thousands of community-built bots overnight. Disney had sent a cease-and-desist in late September 2025, and NBC and DreamWorks reportedly followed.

The community called it the “Moderatedpocalypse,” and the reaction was raw. “I literally was IN THE MIDDLE OF A CHAT with a bot THAT I MADE for ME,” one person wrote on Reddit. Another: “I WAS VERY INVESTED IN EVERY SINGLE ONE OF THOSE CHATS!”

The bot removals were not just a policy change — they were an erasure of creative work that people had spent real time building.

Why Does Character AI Feel So Much Worse Than It Used To?

Character AI does feel worse — and the complaints aren't just vibes. The platform's quality has measurably declined across multiple dimensions, and community frustration is backed by data.

Why is character ai getting worse? Why does character ai suck now? The answer isn't one thing. It's five things happening simultaneously. If filter-related decline is your main pain point, the how to bypass character ai filter guide walks through what actually works (spoiler: not much), and does character ai still have a filter covers the current filter state.

Memory That Forgets and Characters That Drift

The most fundamental complaint: characters forget established details and personalities collapse mid-conversation. This is not a niche gripe.

Academic research analyzing 318 Reddit posts from self-disclosed teen users found that conversational context comprehension was the most common expectation — cited in 61.78% of analyzed posts. People primarily expect their bots to remember their own persona and relationship details.

When we tested Character AI, the memory degraded noticeably after extended sessions — our character forgot details we'd established 20 messages earlier. Has character ai gotten worse at memory specifically? Based on everything we've seen, yes.

Filters That Fire When They Shouldn't

The content filters have overcorrected. Our testing confirmed that filters triggered on emotional roleplay that had nothing to do with explicit content — a grief scene, not anything sexual.

A study of teen users found that 5.4% reduced or quit Character AI specifically because responses became “too censored” after filter updates. And the Change.org petition demanding filter removal has accumulated 174,882 verified signatures as of April 2026 — one of the largest organized protests against a chatbot content policy.

“We believe that the NSFW filter is a hindrance to the full potential of Character.AI and infringes upon the freedom of expression,” the petition states. Is character ai getting worse because of filters? For a lot of people, that's the entire reason.

The Quiet Changes Nobody Talks About

Two things most people don't know about.

First: “Soft Launch” mode, a behavioral dampening layer introduced in May 2025 that controls emotional escalation, introduces hard emotional resets approximately every 8–10 messages, and prevents attachment loops — all without changing the formal NSFW rules. A hidden layer is actively dampening emotional depth. That's the real culprit.

Second: the Charms currency system. Free-tier conversations now get capped mid-thread once a per-thread message threshold is hit, plus a separate 30-swipe daily limit. Full-screen ads interrupt chats mid-conversation. If no-limit chat is what you need, our nsfw ai chat no message limit roundup covers platforms that don't meter you mid-scene.

Character AI's Google Play rating sits at 3.3 stars from over 2 million reviews. “Character AI has officially ruined itself,” one person wrote. “Every update somehow manages to make the site worse... prioritizing monetization over quality.” Another described it as “borderline unusable.”

The grief is real, too. Research on AI companion communities found that 16.73% of all discussions were about grief triggered by model updates; 20.63% were comparative evaluations of whether the AI was “still the same person” after a change.

One in five conversations in these communities is someone trying to figure out if the relationship they cared about survived a software patch. This isn't just complaining. It's been studied and documented in peer-reviewed papers.

Has Anything Actually Improved?

In fairness — some things have. PipSqueak 2, released April 14, 2026, claims improvements to character consistency and memory retention. Chat history expanded from 100 to 1,000 messages. Outages dropped 85% year-over-year. These are real improvements.

But the core complaints remain. Filters still fire on benign content. Memory still drifts after extended sessions. Identity still collapses mid-conversation. Has character ai gotten better at all recently? In spots. The trajectory of the overall experience is still downward for most people, though.

Will the Lawsuits Force Character AI to Shut Down?

The lawsuits are serious — but they're more likely to reshape Character AI than to kill it.

The legal situation is genuinely complex. In May 2025, a U.S. District Judge ruled that Character AI is a “product” subject to product-liability law. That ruling rejected the company's First Amendment defense and let the Megan Garcia wrongful-death lawsuit proceed.

Then in January 2026, Character AI and Google settled five teen-suicide and self-harm lawsuits filed across Florida, New York, Colorado, and Texas. Financial terms were not disclosed. Senators Hawley and Blumenthal have publicly pressed for stronger regulatory action.

And California SB 243, signed October 2025 and effective January 2026, became the first U.S. state law mandating safety safeguards for AI companion chatbots.

The settlements tell the story. Character AI (and Google, as co-defendant) would rather pay and change behavior than fight in court. For the full breakdown of what these incidents mean for everyday users, see our piece on is character ai safe.

CEO Karandeep Anand has said that “safety and regulation in artificial intelligence need to be built into the platforms instead of being an afterthought.” Regulation is moving toward compliance mandates, not bans. SB 243 requires safety safeguards — it does not ban AI companions. The lawsuits will keep shaping what Character AI looks like. They won't end it.

Is Character AI Shutting Down for Good, or Could Things Change?

Character AI is not shutting down for good — but it is becoming a fundamentally different platform than the one people fell in love with.

Financial signals point toward viability, not collapse. Revenue more than doubled from 2023 to 2025 and is projected to nearly double again in 2026. New features like AvatarFX and Scenes signal a pivot from pure chat to multimedia. The company is investing in its future, not winding down.

Will character ai shut down? The money says no. Is character ai shutting down 2025 or at any fixed date? No — that has never been announced and there is no evidence to suggest it.

But here's the tension nobody at Character AI seems willing to address: the platform's 30-day retention rate sits between 13% and 18%. That means 82–87% of people churn within the first month.

The voluntary exodus is the most telling signal — not of a shutdown, but of a platform that's lost the thing that made people stay.

“They do not care about us whatsoever,” one person wrote on Reddit. “I've honestly given up,” wrote another. “I've cut my usage down for the better of my mental health.”

Will Character AI reverse these changes if enough people complain? The 174,882 petition signatures haven't moved the needle. Legal pressure and regulatory mandates make reversal unlikely. Is Character AI going to get even stricter? Almost certainly. The direction is toward more restrictions, not fewer.

What Are the Best Character AI Alternatives Right Now?

The best character ai alternative options right now are ourdream.ai for depth and creative control, Janitor AI for free casual chat, Crushon for visual customization, and Chai for quick mobile conversations. For a full roundup across the category, start with our best nsfw ai chat guide.

Before we break these down: are the alternatives actually better, or just different? It depends on what drove you away.

If filters and memory are your pain points, some alternatives genuinely solve those problems. If you're looking for Character AI's specific vibe with its massive bot library, nothing replicates that exactly. Several platforms do offer what Character AI used to — and in some cases, more.

1. ourdream.ai — Built for Creators Who Want More

If the things driving people away from Character AI are memory drift and filter walls, ourdream.ai was built to address exactly those pain points — and it does.

Four-layer memory system. Auto Memory Log (tracks conversation history automatically), Pinned Memories (things you want the character to always remember), Custom Instructions (behavioral guidelines), and User Personas (switch your own identity per chat).

What that means in practice: your character doesn't forget who they are or who you are mid-conversation. That alone fixes the single biggest complaint about Character AI.

Deeper character customization than anything else in the space. You're building from scratch on the AI girlfriend creator across 46 personality traits, 135 occupations, 64 fetish options, 14+ relationship types, and backstories up to 100,000 characters of narrative depth.

There are also 18 voice options, group chats with up to 12 characters, and 5 AI model choices. And no NSFW restrictions. No filter walls, no mid-scene censorship, no hidden behavioral dampening layers quietly flattening emotional depth. For a broader overview of multimedia chat, see our guide to ai sex chat.

And the scale backs this up: ourdream.ai has over 63 million registered people and 2.1 million monthly active premium subscribers.

Honest cons. There's a learning curve with the character creator if you've never built from scratch before — this is not a “pick a bot and start chatting” platform. It's browser-only with no mobile app, which is a real inconvenience for people who mainly chat on their phones. And the free tier is a 55-dreamcoin one-time trial (roughly 55 messages on the Regular model, not a renewable monthly allocation), so you'll need the premium plan ($9.99/month annual) for sustained use.

But for people who want depth and creative freedom without walls — the things Character AI used to offer — ourdream.ai is the clearest upgrade.

2. Janitor AI — Free Casual Chat

Janitor AI is the straightforward pick for casual chat. It has a large community-built character library, a decent free tier, and it does not hit you with aggressive monetization.

Conversations are accessible and the barrier to entry is low. The cap: memory is basic, and conversations can flatten over time. Solid for browsing existing characters without building anything — but don't expect the depth to hold up across long roleplay arcs.

3. Crushon — Visual Customization

Crushon leans into visual customization. The interface makes character appearance creation accessible, and the platform allows NSFW content.

It's a lightweight option that works if visual design is your priority. Conversation depth doesn't match the visual polish, though — good for creating, less good for sustained storytelling.

4. Chai — Quick Mobile Chat

Chai is mobile-first and casual. Quick conversations, simple character creation, easy to pick up.

Conversations stay surface-level, and the simplicity that makes it accessible also limits what you can build. A decent option for quick mobile sessions; not a replacement for anything deep. If you want a wider category-level pick, our rundown of the best AI girlfriend app options is the next stop.

What Does This Mean for the Future of AI Companions?

Here's something that should give everyone pause: 16.73% of AI companion community discussions, according to peer-reviewed research, are about grief triggered by model updates.

Not frustration. Not annoyance. Grief — the kind usually reserved for loss. When a model changes, the “person” people knew changes too. No Terms of Service clause makes that feel less real.

This is not a Character AI problem exclusively; it's a structural problem with how AI companion platforms work. People invest emotional energy into relationships with systems they do not control. When those systems change underneath them, the loss is genuine even if the servers are still running.

And then there's the tension between safety and freedom that nobody has resolved. California SB 243 exists for real reasons — teenagers have been harmed. The lawsuits exist for real reasons — families lost children. And 174,882 petition signatures exist for real reasons too — people want creative autonomy over experiences they care about.

Both sides have legitimate concerns, and pretending otherwise is dishonest. The question isn't whether to regulate AI companions. It is whether regulation and platform changes can protect people without erasing what made these platforms meaningful in the first place.

FAQ

So Where Does This Leave You?

Character AI is not shutting down. The servers are running, the revenue is growing, and the investment dollars are stacked deep enough to keep the lights on for years.

But the Character AI that people remember — the one where characters felt like creative partners, where conversations could go wherever you took them, where the platform disappeared behind the experience — that version may already be gone. And that loss is real, even if the status page shows 99.48% uptime.

The platforms where that creative spirit still lives — where memory holds, where filters don't interrupt emotional depth — those platforms exist. Places like ourdream.ai were built for exactly the kind of person asking these questions right now.

But whether you stay or go, the larger pattern is worth sitting with. We build creative lives on platforms we do not control. When those platforms change, the grief is genuine; the question of what comes next is one only you can answer.

Sometimes a platform doesn't have to shut down to stop being the place you want to be.