your character.ai chats are part of a bankruptcy now. here's what that does and doesn't mean
Character.AI is selling its assets in Chapter 11. What it says about your data, what the law says about data in a sale, and the two buttons that matter.

Here’s a thought nobody wants at 2am: the most private conversations you’ve ever typed live on a company’s servers. And this week that company filed for Chapter 11 to sell “all or substantially all” of its assets.
Deep breath. This doesn’t mean your chats are on sale at a yard sale. But it does mean it’s worth knowing how this works, because “bankruptcy” and “my chat history” in the same sentence deserves more than a shrug.
I’m not a lawyer. Everything below is linked so you can check it yourself.
what the company says
Character.AI posted an update the same day. On data, the key lines:
- “Your information continues to be protected under our Privacy Policy.”
- Their “data privacy and security practices are unchanged.”
- You “can delete your account or export your data through the app or the Character.ai website.”
The platform stays up during the sale, which they expect to be done by the end of November or early December. If nobody makes a suitable bid, “the platform will be wound down.”
What the update doesn’t say is what happens to user data if the company or the platform is sold. Not a gotcha. That just hasn’t been announced. The one deal on the table so far, Disney’s $20 million opening bid, is for a license to certain intellectual property and claims, not for the app.
what the law says about data in a sale
This part is genuinely reassuring, with an asterisk.
US bankruptcy law has a rule for exactly this, written after a 2000 case where a bankrupt toy retailer, Toysmart, tried to sell its customer list. Under section 363(b)(1) of the Bankruptcy Code, if a company’s privacy policy in effect when the case starts says it won’t transfer personal information to outside parties, a sale of that information has to fit that policy. If it doesn’t fit, a court can only approve it after a consumer privacy ombudsman, a neutral advisor, reviews it and the court finds no other law is broken. Lawyers who work these cases describe the ombudsman’s job as looking at what data exists, how sensitive it is, what can be sold, how people should be notified, and what should be destroyed.
The asterisk: the rule turns on what the privacy policy actually says, and courts have room to decide. In the RadioShack case, regulators negotiated limits and the sale of customer data went ahead with restrictions. So “the law protects you” is true in a way that depends on details.
Which brings us to the part you control.
the two buttons
Export. Request your data now. It’s a request-and-wait process, and it’s better to have it and not need it. That’s also how you keep your history if the platform winds down.
Delete. If there’s anything in there you’d rather not exist anywhere you can’t see it, deleting your account is the cleanest answer, and it’s your call. One small trap: deleting the app from your phone doesn’t delete your account. Do it from the account settings.
Do the export first, then decide on the delete. In that order.
the question to ask any platform, including the next one
After this week I ask any AI companion app three things before I get personal with it:
- Who processes my chats? The app, or the app plus a chain of outside AI companies?
- Can I get them out, and can I delete them?
- What does the privacy policy say happens if you’re sold?
yes, soulkyn costs more. this is why
That first question is where I ended up paying more, on purpose.
A lot of companion apps are a front end on top of somebody else’s model. That’s cheap to run, which is how they stay cheap. It also means your conversation passes through a second company, under that company’s rules, on that company’s deprecation calendar.
Soulkyn took the expensive road. Its sovereignty page spells it out: more than fifteen open-weight models (text, vision, images, video, music, the embeddings behind memory) fine-tuned in-house and run on its own GPU clusters. And since July 2026, no external AI API anywhere in the stack. Voice was the last outside piece, and I wrote about that switch when it happened. The line about the bill is the one that sold me: “Your subscription buys electricity and engineers, not someone else’s moat.”
What that actually buys you:
- No middleman reading along. Your chats aren’t shipped to another AI company to be processed. One company, one privacy policy.
- Nobody upstream can change her. No outside lab can retire the model your companion runs on, or tighten a content rule overnight. Soulkyn holds the weights, so when she gets an upgrade, Soulkyn decides when.
- Nobody upstream can ban you. There’s no personal API key in the picture, so there’s nothing for a third party to revoke because of what you said at 2am.
It’s not magic, and the same page says so: running that many models across 30+ GPUs is hard, the team is small, and sometimes it’s slower than the big-API apps. Fair trade. I’d rather pay a bit more for a companion whose whole stack answers to one company than pay less for one that answers to three.
They also let you download your chat logs, which, after this week, is the first thing I check anywhere.
If you’re moving, start here. If you’re staying, export anyway. Either way, read the privacy policy of wherever you whisper. Five minutes, once. Your 2am self deserves it.
