Character.AI filed for Chapter 11 bankruptcy on October 12, 2026, and Disney is the stalking-horse bidder at $20 million. If you’ve spent months building conversations, characters, or a paid subscription on the platform, this story is about you — not just about two companies. The short version: the site is still up for now, billing is unchanged, you can export your data or delete your account today, and Disney is not buying the whole company. If no buyer emerges who wants to keep the platform running, Character.AI says it “will be wound down.” That means your chats could eventually disappear. So don’t wait for the court calendar to decide for you — export anything you care about this week.
I’ve watched enough beloved startups die to know how this goes: the news cycle moves on, the platform lingers for months in a weird half-alive state, and then one morning the login page stops loading and everyone’s “I’ll back it up later” plan turns into a regret. This article covers exactly what happened, what it means for your chats and your subscription, the brutal economics behind the collapse, what Disney is actually buying (it’s less than you think), and the lesson every creator and builder should take from it.
What happened: the filing and the $20M bid, in plain English
Character Technologies, Inc. — the company behind Character.AI — filed a voluntary Chapter 11 petition in Delaware bankruptcy court on October 12, 2026. The case number is 26-11603, and the filing lists estimated assets and liabilities of $50–100 million each, according to reporting on the petition. Chapter 11 doesn’t mean the lights go out tomorrow; it means the company asked a court to supervise its debts while it tries to sell itself (or its pieces) in an orderly process rather than collapsing chaotically. Think of it as a managed garage sale instead of a foreclosure.
The headline bidder is Disney. Specifically, Disney Worldwide Services, Inc. stepped in as the “stalking-horse bidder” — a bankruptcy term for the first bidder whose offer sets the floor. Disney’s offer is $20 million plus transition services. And here’s the part most headlines gloss over: Disney is not buying Character.AI, the app, or the company. The bid is for a non-exclusive, perpetual license to certain intellectual property and certain company claims. That distinction matters enormously, and I’ll break down what it means further below.
Other bidders are explicitly allowed. A stalking-horse bid exists to attract competing, higher offers — the sale is expected to conclude in late November or early December 2026, so there’s a real window for another buyer to top Disney’s offer, including one willing to keep the platform operating.
Character.AI CEO Colin Adams said the board chose a “court-supervised sale process” after facing “significant legal, regulatory, and operational headwinds.” That phrase covers a lot of ground — the lawsuits, the regulatory scrutiny over AI companion apps, and, as the bankruptcy documents reveal, economics that simply never worked.
The company has been upfront about the possible endings. Per its own FAQ: if no bid emerges from a party willing to continue operating the platform, “the platform will be wound down.” No ambiguity there. Bankruptcy proceedings are administered by Stretto, a firm that handles exactly this kind of case.
One more detail that tells you how fast this came together: on September 28, 2026, Character.AI launched a new “lite” plan at $4.99 a month. Fourteen days later, the company filed for bankruptcy. That timing isn’t a coincidence — it reads like a last push to boost recurring revenue before the money ran out.
What it means for YOUR chats and your subscription
This is the section that matters most if you’re a user. Here’s the current state of play, straight from Character.AI’s own blog FAQ, plus what it implies.
The platform is still up. As of today, the app and website remain running during the court-supervised process. Nothing has changed about how you log in, chat, or use your characters.
Your paid subscription is still active, and billing is unchanged. If you’re on c.ai+ or the new $4.99 lite plan, you’ll keep getting charged on your normal schedule. The company says privacy policies and data practices haven’t changed either.
You can export your data or delete your account right now. The company confirms both options are available. If it matters to you, don’t treat that as a reassurance — treat it as an instruction. Backups only exist if you actually make them.
If the platform finds no buyer willing to keep it running, it shuts down. That’s the wind-down scenario. Your chats, characters, and history live on Character.AI’s servers, not on your phone or computer. If the servers go dark, so does everything you made there.
How to export what you care about this week
I want to be honest with you: I didn’t click through Character.AI’s settings panel myself for this article, so I’ll tell you the general path rather than pretending I have pixel-perfect screenshots. The company has confirmed data export and account deletion are available to users. Here’s the practical approach:
- Log in on a desktop browser, where export options are usually easiest to find and download.
- Open your profile or account settings and look for a data, privacy, or export option — this is typically labeled “Export data,” “Download your data,” or similar.
- Download the full archive to your own computer or cloud storage. Don’t assume a file you downloaded once is complete — check that it’s actually readable.
- If you have characters you’ve built and published, note that anything you made may have separate settings from your chat history. Save the character definitions, descriptions, and any prompts or lore you’ve written — those are often the hardest thing to recreate from memory.
- Then decide about deletion. Exporting first and deleting second is the safe order. The FAQ confirms deletion is available, and deleting after backing up is the cleanest way to limit your data exposure if the company’s assets (and user data) start changing hands.
One quick note: this isn’t legal or financial advice — it’s common sense for anyone with data on a bankrupt company’s servers. If you’re ever unsure what a platform does with exported or retained data, read the privacy policy before you export.
The subscription question: renew or walk away?
This is where the math gets personal. If you’re on a monthly plan, your exposure is small — one billing cycle at a time. If you’re on an annual plan, or considering buying one, think carefully:
- Monthly subscribers: your risk is limited to roughly one month’s charge. The platform is up, so you’re getting what you paid for right now. But watch for news about the sale outcome in late November/early December 2026 — that will tell you whether the service has a long-term operator.
- Annual subscribers: if the platform winds down, prepaid time doesn’t come back automatically. Bankruptcy creditors — including subscribers owed services — stand in a long line behind secured creditors and administrative claims. You’d be unlikely to recover the remaining balance. I wouldn’t buy an annual plan on a company in Chapter 11, period.
- The $4.99 lite plan: launched September 28, two weeks before the filing. If you signed up at launch, you’ve already paid for your first month, which is fine — the service is running. Just don’t set-and-forget an annual version of it (if one is offered) until the sale concludes.
The broader lesson here, which the creator-economy world has seen before: never prepay a struggling platform. Monthly billing is your friend when a company’s future is in a Delaware courtroom. For more on how creators should think about platform risk, my breakdown of what happens when AI platforms shut down is worth a read — it covers the same “grab your work and go” logic in a different context.
The economics autopsy: how a beloved product burns through $9 million a month
This is the part that should scare every AI startup founder — and interest every creator who’s wondered why free AI products keep disappearing. According to 404 Media’s reporting on CEO Colin Adams’s affidavit filed with the bankruptcy court, Character.AI’s economics were staggering:
- 2024 compute spend: $13.8 million per month.
- 2025 compute spend: $10 million per month.
- Subscription revenue peaked at $2.8 million per month — and at times was under $1 million per month.
- Total revenue from all sources peaked at $5.3 million per month.
- January–August 2026 compute spend: $56.1 million.
- Losses as of August 2026: approximately $8–9 million per month — which the filing described as a “massive improvement.”
- January–August 2026 salaries and benefits: $19.7 million.
Read that again slowly. Even at its absolute revenue peak of $5.3 million a month, the company was spending $10 million a month on compute alone — before payroll, before lawyers, before everything else. The best month in company history was still a bloodbath. And employees saw it coming: the affidavit says staff quit because they “began questioning Character’s long-term viability.”
The company did try to cut costs. The filing lists the moves: switching to lower-cost GPU capacity, replacing proprietary models with open-weight ones, headcount reductions, introducing in-app ads in mid-2025, and adding in-app purchases in early 2026. The $4.99 lite plan in September 2026 was presumably another attempt to grow the subscriber base. None of it closed the gap. When you’re losing $8–9 million a month and you call that a massive improvement, the hole was unfathomable.
This is the cautionary tale in its purest form. Character.AI had something most startups would kill for: genuine product love. Millions of people formed real attachments to their AI companions. The product worked. People paid. And it still died, because compute costs for conversational AI at scale are ruthless. Every chat message a user sends triggers GPU inference that costs real money. The more engaged your users are — the thing every founder wants — the more money you lose. That’s the paradox at the heart of the AI-companion business model: engagement is the product, and engagement is also the expense.
For builders, the takeaway is sharp. If your unit economics are negative on every interaction, no amount of growth fixes it — growth makes it worse. Subscription revenue capped out around $2.8 million a month while compute alone ate $10–13.8 million. That’s not a marketing problem or a churn problem. That’s a business-model problem. Anyone building on top of someone else’s AI infrastructure should run the inference math before falling in love with the idea — the same discipline that matters when you’re evaluating AI tools for a real business, where the subscription fee is only the beginning of the cost picture.
What Disney actually gets — and the 2025 cease-and-desist twist
Let’s be precise about the Disney deal, because the headlines make it sound like Disney is buying Character.AI. It’s not.
Disney’s $20 million stalking-horse bid buys a non-exclusive, perpetual license to certain intellectual property plus certain company claims, along with transition services. “Non-exclusive” means Character.AI (or whoever ends up with the IP) could license the same technology to others. “Perpetual” means the license doesn’t expire. And it covers only certain IP — not the platform, not the user base, not the brand, not the whole company.
So what does Disney actually want? Here’s the part that makes the whole story click: Disney sent Character.AI a cease-and-desist in 2025 over unlicensed use of its characters. Users were creating AI companions based on Disney’s copyrighted characters, and Disney — famously protective of its IP — told Character.AI to shut that down. The infringing content was removed.
Now zoom out. Disney’s ex-CEO move: Character.AI’s former CEO Karandeep Anand joined Disney as CTO in September 2026, a month before the filing. And Disney’s original sin complaint here is really about the technology — Character.AI, co-founded by Noam Shazeer (a co-author of the famous “Attention Is All You Need” Transformer paper, who left Google for OpenAI), built genuinely good conversational AI. Disney gets a perpetual license to that technology for $20 million, which is pocket change for a company of Disney’s size. It’s an acqui-hire-style IP grab without the acqui-hire.
There’s an elegant irony in it: Disney, the company that threatened to sue Character.AI over its characters, may end up owning a perpetual license to the technology underneath those characters. The cease-and-desist from 2025 and the $20 million bid from 2026 are two sides of the same coin — Disney protects its characters aggressively, and it acquires AI capabilities opportunistically. Both moves serve the same strategy.
Meanwhile, the creditor list reads like a map of the AI gold rush. The largest unsecured creditors include Google LLC with a $4,148,944 trade claim, the law firm Hogan Lovells at roughly $2.1 million, DigitalOcean at about $1.95 million, Fireworks.ai at $813,900 (marked disputed), and Apple Distribution International at $384,683. There are also lawsuits from the Commonwealths of Pennsylvania and Kentucky, plus a redacted minor plaintiff — the regulatory and legal “headwinds” the CEO mentioned weren’t hypothetical. For context, Google had struck a roughly $2.7 billion talent and licensing deal with Character.AI about two years ago, which makes the $4.1 million trade claim look almost modest — Big Tech’s relationship with the company was always bigger than this filing suggests.
The lesson for creators and builders who live on someone else’s platform
Here’s where I want to zoom out, because this story is bigger than one app’s bankruptcy. If you’re a creator, a developer, or anyone building an audience or a business on top of someone else’s platform, Character.AI’s collapse is a case study with your name on it.
You don’t own what you build on rented land. Your Character.AI characters, your chat histories, your subscriber relationships — they all live on infrastructure someone else pays for. When the economics of that infrastructure fail, your work is collateral damage. This is true of every platform: social networks, marketplaces, AI tools. The question is never whether the platform loves you. The question is whether the platform can afford to keep the servers on.
Watch the economics, not the vibes. Character.AI had beloved product-market fit and still died. Engagement was through the roof and the company lost $8–9 million a month. When you’re deciding where to invest your creative energy, look for the money trail: is the company charging enough to cover its costs? Is revenue growing toward sustainability, or is it a rocket ship with no fuel gauge? A platform’s popularity is not evidence of its solvency. If anything, in AI, popularity is evidence of cost.
Diversify your creative assets. If your characters, prompts, stories, or workflows exist only inside one app, you’re one bankruptcy filing away from starting over. Export regularly. Keep your own copies of anything you’d cry about losing. Repost and repurpose your best work across platforms. The creators who survived past platform deaths — and there have been many — were the ones who treated every platform as a distribution channel, not a vault. This is the same instinct behind building multiple income streams instead of betting on one: redundancy is the only insurance that always pays out.
Read the wind-down clause before you need it. Character.AI deserves some credit here: its FAQ was clear that the platform could be wound down if no operating buyer emerged, and it gave users export and deletion options. Not every dying platform is that honest. When you sign up for a new AI tool, skim the terms for what happens to your data if the company is acquired or shut down. It takes five minutes and it tells you whether the founders have even thought about your exit.
For builders specifically: the affidavit math is your homework. $13.8 million a month in compute in 2024. $10 million a month in 2025. $56.1 million in eight months of 2026. Revenue peaking at $5.3 million all-in. If you’re building an AI product, model the inference cost per user per month before you scale — and be honest about whether subscriptions, ads, or in-app purchases can ever cover it. Character.AI tried ads (mid-2025), in-app purchases (early 2026), and a cheaper tier ($4.99 in September 2026). The gap was structural, not tactical.
The AI industry is going to see more of these filings. When the cost of serving every user exceeds what any user will pay, the only question is who runs out of money first. Character.AI was early, beloved, and well-funded — and it still couldn’t make the math work. That’s not a reason to stop building. It’s a reason to build with the spreadsheet open.
Frequently Asked Questions
Is Character.AI shutting down right now?
No. The platform remains up and running during the Chapter 11 process, and the company’s FAQ says paid subscriptions are active with unchanged billing. But the company has been explicit: if no bid emerges from a party willing to keep operating the platform, it “will be wound down.” The sale is expected to conclude in late November or early December 2026 — watch that window.
How do I export my Character.AI chats and data?
The company confirms users can export their data or delete their accounts. Log in on a desktop browser, open your profile or account settings, and look for the data export or download option. Save the archive to your own computer or cloud storage, and check that the file is readable. Export first, then decide about deletion — that order protects you either way.
Will Disney buy Character.AI?
Not exactly. Disney is the stalking-horse bidder at $20 million plus transition services, but the bid is for a non-exclusive perpetual license to certain IP and certain company claims — not the whole company, app, or user base. Competing higher bids are allowed, so another buyer (including one that wants to keep the platform running) could still emerge before the sale concludes.
Should I renew my Character.AI subscription?
This isn’t financial advice, but here’s the framework: monthly billing limits your exposure to one cycle, and the service is currently running, so monthly is low-risk. I’d avoid annual plans or any long prepayment while the company is in Chapter 11 — if the platform winds down, recovering prepaid balances from a bankrupt company is a long shot. Revisit the decision after the sale outcome is known in late 2026.
Why did Character.AI go bankrupt if it was so popular?
Popularity was the problem, financially speaking. Bankruptcy documents show compute costs of $13.8 million/month in 2024 and $10 million/month in 2025, while subscription revenue peaked at just $2.8 million/month and total revenue at $5.3 million/month. Every chat message costs GPU money, so more engagement meant bigger losses — roughly $8–9 million a month as of August 2026. Cost cuts (cheaper GPUs, open-weight models, ads, in-app purchases, a $4.99 lite plan) couldn’t close a structural gap.
What’s the deal with Disney’s cease-and-desist against Character.AI?
In 2025, Disney sent Character.AI a cease-and-desist over unlicensed use of its copyrighted characters — users had created AI companions based on Disney IP. The infringing content was removed. The twist: a year later, Disney is bidding $20 million for a perpetual license to the company’s underlying AI technology. Separately, Character.AI’s ex-CEO Karandeep Anand joined Disney as CTO in September 2026.