What Happens to User Data After Character.AI's Bankruptcy?

The news that AI companion platform Character.AI has entered bankruptcy proceedings is pushing a type of asset that has rarely been seriously discussed before into the spotlight: the long-term conversation data accumulated between users and fictional characters. New York Attorney General Letitia James, joined by as many as 40 bipartisan state attorneys general, has filed a notice of interest with the bankruptcy court, with one core demand — before the data is sold as an asset, regulators must first weigh in on consumer protection issues.

Why This Bankruptcy Case Is Especially Sensitive

When an ordinary company goes bankrupt, creditors typically care about equipment, patents, and accounts receivable. But Character.AI's core asset composition is entirely different: its value derives largely from the emotional outpourings, psychological state descriptions, interpersonal relationship details, and even usage records of minors that users continuously input.

What makes this data special is that it is not the kind of routine information users "rationally submit" in exchange for a service, but content that flows naturally through the process of forming an emotional bond with an anthropomorphized character. Many users treat Character.AI as a confidant, discussing highly private topics involving depression, loneliness, and family conflicts.

Because of this, when bankruptcy proceedings could potentially package and sell this data to an unknown third-party buyer, the nature of the issue shifts from "business reorganization" to "how to handle the privacy of hundreds of thousands or even millions of people."

What the Joint Action by 40 State Attorneys General Means

According to information disclosed by Letitia James's office, this is not an isolated statement from a single state. Forty attorneys general from both parties jointly filed a notice of interest, explicitly stating their intent to make all parties involved in the bankruptcy case "aware of the regulatory issues involved."

The legal significance of this move is that it does not directly halt the bankruptcy proceedings, but reserves space for potential future intervention. The attorneys general are attempting to embed consumer protection requirements before the transaction structure for the data sale is finalized — for example, requiring the buyer to accept specific data usage restrictions, or requiring users to be granted informed consent and choice before data transfer.

The multi-state joint intervention also reflects a consensus: the regulatory归属 of user data accumulated by AI companies cannot simply follow the logic of "asset liquidation" under traditional bankruptcy law.

Notice of Appearance / Statement of Interest is a formal legal document in U.S. bankruptcy proceedings. Creditors, regulatory agencies, or other interested parties can file such a notice with the bankruptcy court to indicate that they have an interest in the case, thereby obtaining service of case developments and gaining procedural standing to express opinions at key junctures (such as asset sale motions or reorganization plan votes). It is not itself a formal lawsuit or intervention, but rather a "stake your claim first, act later" strategy: once the court or debtor advances a transaction involving data sales, the filing party can quickly raise objections or file conditional applications. When multiple state attorneys general jointly file the same notice of interest, it typically means they have coordinated their positions and are prepared to intervene as amicus curiae or direct parties when necessary to impose regulatory conditions on the transaction structure. This is one of the most commonly used preliminary measures by state-level law enforcement agencies to protect their state's consumer interests in federal bankruptcy proceedings.

What Typically Happens to Data in Bankruptcy

Under standard bankruptcy procedures, company assets are inventoried, valued, and then transferred to buyers through auction or negotiation, with proceeds used to repay debts. User data in this framework is often classified as "intangible assets" or "customer relationships."

The problem is that existing law is unclear on how to characterize "AI conversation data." It can be understood as business records or as personal privacy information, and different characterizations lead to completely different handling rules. If treated as ordinary commercial assets, the data may transfer automatically with the transaction; if deemed specially protected consumer data, the transfer would require additional conditions.

Character.AI's user agreements typically contain data usage and data transfer clauses, but whether such clauses can effectively bind buyers in a bankruptcy context is itself a question that has not been fully tested in court.

A particularly noteworthy mechanism in bankruptcy law is the "Section 363 sale." Section 363 of the U.S. Bankruptcy Code allows a debtor, with court approval, to sell all or part of its assets to a buyer before a reorganization plan is confirmed — commonly seen in cases where time is tight or asset depreciation risk is high. A key feature of such sales is "free and clear": the buyer can typically acquire the assets without inheriting the seller's debts and liabilities. If user data is included in a 363 sale, the buyer may argue that it does not assume obligations under the original platform's privacy policy, and users' deletion requests or usage restriction commitments may become void. This is precisely why regulators often raise objections at the 363 sale motion stage, demanding that the court attach data protection conditions to the sale order or require the buyer to formally assume specific privacy commitments in writing. This explains why the 40 attorneys general's notice of interest needed to intervene early, before the transaction structure was determined.

The Actual Risks for Users

From the user's perspective, the most direct risk is the transfer of data control. The company that originally promised you specific privacy protections may no longer exist, and the buyer taking over the data is not bound by the original commitments — or operates only within the minimum legal requirements.

Second is the change in data usage. Data used for training and improving AI models could be used for entirely different commercial purposes, and users would neither know about it nor be able to object.

Third is the failure of deletion rights. In bankruptcy proceedings, the company may no longer have the resources or incentive to handle user deletion requests, and once data is transferred to a new entity, the difficulty of tracing and deletion further increases.

The Bigger Picture

The Character.AI case is not an isolated event. As a wave of AI startups enters a stage where they must face financial realities, the question of user data ownership and disposal will recur. The current legal framework has obvious gaps when dealing with "emotional AI interaction data," and existing privacy regulations may not directly cover data transfers in bankruptcy scenarios.

The action led by New York State is, to some extent, setting a precedent for future cases: when an AI company's user data becomes a bankruptcy asset, regulators have both the authority and the will to intervene. How this ultimately plays out will affect not just Character.AI's users, but the entire AI consumer product industry's expectations around data governance.

For ordinary users, the practical reminder this provides is: when engaging in deep interactions with AI characters, it's worth maintaining a clear-headed awareness of the platform's long-term stability and data policies — what you input may one day become a line item on a balance sheet.

One-line Summary

Character.AI's bankruptcy case has drawn joint intervention from 40 state attorneys general, focusing on whether AI emotional conversation data can be sold as a bankruptcy asset.

Paragraph Summary

After Character.AI entered bankruptcy proceedings, its core asset — the deep emotional conversation data accumulated between users and AI characters — became the focus of contention among all parties. New York Attorney General Letitia James, joined by 40 bipartisan state attorneys general, filed a notice of interest demanding that the bankruptcy court incorporate consumer protection considerations before any data sale. This data involves highly private content such as mental health and family conflicts, and includes records of minors, making its disposal far more sensitive than ordinary commercial data. Existing law is ambiguous on how to characterize AI conversation data in bankruptcy, and whether user agreements can bind buyers remains a judicial gap. Users face threefold risks: transfer of data control, changed usage purposes, and failed deletion rights. This case may set a precedent for data governance when AI companies go bankrupt, affecting the entire industry's data policy expectations.