What Happens to Your Data When an AI Company Fails?

AI conversations can contain information people would hesitate to share anywhere else: relationship problems, health concerns, financial worries or confidential business plans. Yet the company storing those conversations may eventually be acquired, restructure or close.

Character.ai’s bankruptcy brings that risk into focus. The company announced a voluntary Chapter 11 process in October 2026 to pursue a court-supervised sale. Its platform remains operational, but the company says it could wind down if no proposal supports continued operations.

For businesses adopting AI, the case raises a practical question: what happens to sensitive information when the provider’s business changes?

Private conversations can become part of a transaction

Character.ai’s proposed transaction with Disney involves a $20 million offer for certain assets, including a non-exclusive licence to intellectual property. Disney’s offer does not include the platform or user data. The wider sale process, however, has raised separate concerns about the future of users’ conversations.

State attorneys general have intervened to seek protections for sensitive consumer information, including children’s data. Their concerns reflect the nature of the material involved: conversations about mental health, relationships and emotional struggles.

These records carry consequences beyond ordinary account information. A conversation may reveal intimate details about its author, but also about a partner, colleague, customer or family member who never used the service.

For a business, the equivalent could be an employee pasting a customer complaint, an internal investigation or a commercial proposal into an AI assistant. The organisation’s exposure then extends beyond the person who created the account.

A provider’s failure changes the privacy context

Bankruptcy does not automatically make personal information available to any buyer, nor does it remove every privacy obligation. What can happen depends on applicable law, contractual commitments, privacy policies and the court-supervised process.

The uncertainty itself matters. Users may have chosen a service because they trusted its brand, management or stated purpose. A new owner may have different commercial priorities, even where existing obligations continue to apply.

The distinction between an AI assistant’s responses and the provider’s actual commitments is also critical. A chatbot’s reassuring statement that a conversation is confidential cannot substitute for a contract, privacy policy or enforceable data protection controls.

AI adoption needs an exit plan

Businesses often assess AI tools through capability, speed and cost. Provider continuity and data handling deserve the same attention.

Before sensitive information enters an AI workflow, teams should establish:

  • What the provider retains: prompts, uploaded documents, outputs, conversation histories and operational logs.
  • How that information may be used: including model improvement, service delivery and access by subprocessors.
  • What happens during a business transfer: including acquisition, restructuring or a sale of assets.
  • How the organisation can leave: through usable exports, defined deletion processes and an alternative operating approach.

These questions should influence the system’s design. An application that sends only the information required for a task limits exposure. Keeping essential business records in systems the organisation controls also makes it easier to replace an AI provider without losing operational continuity.

Where practical, integrations should allow a model or service to be changed without rebuilding the entire workflow.

Deletion helps, but its scope matters

Character.ai says users can export their information or delete their accounts through its app or website. Its privacy guidance also describes rights that vary by region, including access, deletion and restrictions on processing.

For businesses, deleting an account should be part of a documented offboarding process. Teams need to understand what the request covers, whether retention exceptions apply and how completion can be confirmed.

At Control F5 Software, we consider these dependencies part of building software that can be maintained over time. Integrating AI responsibly means planning for changes in the technology, the provider and the business behind it.

An AI service may become part of daily operations quickly. The ability to protect information and move away from that service needs to be designed just as deliberately.

Source

Control F5 Team
Blog Editor
OUR WORK
Case studies

We have helped 20+ companies in industries like Finance, Transportation, Health, Tourism, Events, Education, Sports.

READY TO DO THIS
Let’s build something together