Does the Character.AI bankruptcy put your chats at risk?

What happened in the Character.AI bankruptcy?

Character Technologies, the company behind Character.AI, filed for Chapter 11 bankruptcy in Delaware on October 12, 2026. Disney is the opening bidder at $20 million. Two years earlier, the startup had struck a licensing deal with Google worth roughly $2.7 billion, and Google rehired its founders as part of it. In the filing, the company put both its assets and its liabilities at somewhere between $50 million and $100 million. The CEO blamed significant legal, regulatory and operational headwinds. In plain terms, that means lawsuits over harm to minors, huge server bills and a run of leadership departures. The roughly $190 million the company raised from venture investors didn't last.

The Disney deal is narrower than the headlines suggest. Disney isn't buying the app. Its bid is for a non-exclusive, perpetual license to certain intellectual property and claims. Disney is the stalking horse bidder, which means its offer sets the minimum price. Rival buyers can still come in with a better offer, and a judge has to approve whatever deal wins. The company expects the sale to wrap up in late November or early December 2026. Disney has also hired former Character.AI staff to provide transition services so the platform can keep running for a while. There's some irony here too: in September 2025, Disney sent Character.AI a cease-and-desist letter over its copyrighted characters.

For now, the app is still running and paid subscriptions are still active. The company says it plans to keep the platform going with as little disruption as possible during the sale. If no buyer wants to keep it alive, the platform will be wound down. On privacy, the company says user data continues to be protected under its privacy policy. That policy allows the data it collects, including chat transcripts, voice recordings, images and videos, to be disclosed as part of a sale or sale negotiations. It also reminded users that they can export or delete their data through the app or website. That last detail matters to you even if you've never chatted with a fictional character in your life.

What happens to your chats when an AI company goes bankrupt?

When an AI company goes bankrupt, your chats usually count as company property, and the company's own privacy policy and terms largely decide where they go. Most consumer apps include a business transfer clause, which says your data can pass to a buyer in a merger, an acquisition or a sale of assets. Few people read that clause when they sign up. In bankruptcy, it suddenly matters more than anything else in the document. A US bankruptcy court can also weigh in. When a proposed sale clashes with promises in a privacy policy, the court can appoint a consumer privacy ombudsman to review the deal before it goes through.

We've seen this happen before. When the genetic testing company 23andMe filed for bankruptcy in March 2025, millions of customers suddenly wondered who would own their DNA data. Many rushed to delete their accounts. A court-supervised auction followed, and a privacy ombudsman was involved. The data ended up with a nonprofit led by the company's co-founder for $305 million, after a drug company had placed the earlier winning bid. The lesson isn't that buyers are villains. It's that your data becomes something to sell the day the company runs out of money. If you want the longer version, we've broken down what usually happens to your data when an AI company fails.

In practice, there are three ways this tends to end. In the best case, a buyer keeps the service running under the same or similar terms, and most users barely notice. In the second, your data moves to a new owner whose privacy policy you never agreed to, and they can usually change it later with notice. In the third, the service shuts down and the data is deleted, or it sits in an archive nobody looks after. Here's our view: a clean shutdown is often the least worrying outcome. The quiet sale is the one to watch, because a log of your private conversations with an AI says far more about you than your shopping history.

Why does this matter more for an AI assistant with memory?

It matters more because an AI assistant with memory keeps a growing record of your real life and work, not just a pile of separate chats. On Character.AI, a lot of people were writing stories with made-up characters. That's personal, but a lot of it is fiction. A personal assistant is different. You tell it your accountant's name, your kids' pickup times, which client pays late, what you're charging for the next job and when your lease renews. Over months, that adds up to a fairly complete picture of how you run your week. That's exactly why it's useful, and it's also why you should know who could end up owning it.

Think about the messages you'd actually send. "Remember that Dana's company pays invoices on 60-day terms." "Remind me every Monday at 9 to call the accountant." "Here's the supplier contract, what's the notice period?" Each one is harmless on its own. Together, they include other people's details: client names, phone numbers, prices and contract terms. Those people never agreed to your choice of AI tool. If you work in the EU, you may have your own data protection duties for client information you pass to an outside service. A bankruptcy sale doesn't create that problem. It just makes it obvious, which is a good reason to think about it before it happens.

Here's the contrarian part: the answer isn't to avoid assistants that remember things. An assistant that forgets everything is just a search box you have to keep reminding. You end up re-explaining your clients, your schedule and your preferences every time, and the delegation you were promised never really happens. The better approach is to pick an assistant on purpose, check its terms and decide what you're comfortable sharing. If you want help that actually remembers your context and recalls it when you ask, memory is a feature worth having. Just treat it like hiring a person: check who they work for before you give them the office keys.

What should you check before trusting an AI assistant with your data?

Before you trust an AI assistant with your data, check five things: who owns what you share, whether you can export it, whether you can delete it, what happens in a sale and whether your conversations are used for training. Start with ownership. Read the terms for the line on user content. You want wording that says you keep ownership of what you send, and that the company gets only the license it needs to run the service. Then look for a way to export your data. It doesn't need to be pretty, as long as you can get your stuff out if things go wrong. Character.AI users who could export their chats had options this week. Users who couldn't were left waiting on a court.

Next, look at deletion and the sale clause, because the real risk sits there. Find out whether deleting your account removes your data or just hides it, and how long backups and legal copies are kept. Then search the privacy policy for words like "merger," "acquisition," "sale of assets" or "business transfer." Almost every company has that clause, so the question isn't whether it exists. What you want to know is whether a buyer has to honor the current policy and whether you'll be told before your data moves. After that, check whether your conversations are used to train models, and whether you can opt out. A clear yes or no is fine. Vague wording is a warning sign.

The last two checks are less about the terms and more about the company itself. Find out who's behind it and how it makes money. A free assistant with no visible business model is paying its server bills some other way, and Character.AI shows how fast those bills can grow. Then look at what the assistant can actually do on your behalf: read documents, message people, edit your calendar. The more it can do, the more its permissions matter. We've written a plain-English checklist on what an AI assistant should and shouldn't do without your approval if you want to look at that side in more detail. None of this takes more than twenty minutes, and it's worth doing once.

How do you use an AI assistant safely if the company could go bust?

To use an AI assistant safely, treat it like a sharp personal assistant, not a filing cabinet. Let it do the work, but keep the originals somewhere you control. Contracts, invoices and client lists should live in your own storage. You send a copy to the assistant to get it read, summarized or checked. Share only what the task needs. "Remind me to chase the Levi invoice on Thursday" works fine without the client's ID number or bank details. Every few months, ask what the assistant remembers about you, and trim anything that's no longer needed. The habits are dull, but they're also why the next bankruptcy headline won't worry you.

This is how we think about it at Agent Sonic. You message Sonic in WhatsApp like any other contact, with nothing to install. It keeps a persistent memory so you don't have to repeat yourself, and you can ask it straight out: "What do you remember about the Cohen project?" When Sonic adds a meeting to your Google Calendar or updates a row in Google Sheets, that data lives in your own Google account, not in a system you can't reach. If you connect Notion, Sonic gets read-only access, so it can look things up without changing anything. That's a sensible setup: the assistant does the admin, and your important records stay with you.

Run the same checklist on Sonic that you'd run on anyone else. Read the terms, ask your questions, and only share client details once you're comfortable with the answers. We'd rather you choose us with your eyes open than out of habit. Our honest take on the Character.AI story is that panic-deleting every AI account you own is the wrong lesson. The right lesson is to be careful and deliberate about what you share. Pick tools that let you get your data out, keep your originals, and give the assistant only what it needs to take the busywork off your plate. If that sounds like what you want, get access to Sonic here: leave your details and you'll get the number and a quick onboarding.

Frequently asked questions