Users report AI mistakes that delete accounts and data

In short: A growing set of real-world cases shows AI systems wrongly deleting accounts or data, and recovery often requires going back through automated AI-run processes.

What's going on

Reports have described people losing access to online accounts, or companies losing important records, after automated systems made the wrong call. In several cases, users said there was no clear human to talk to, and the only way to fix the problem was to use the same company’s automated review or recovery tools.

One high-profile example came from Meta, where some Facebook and Instagram users said their accounts were deleted “without cause.” The New York Times reported that people trying to reverse the decision had to rely on Meta’s appeal process, which is largely form-based and automated.

Other recent incidents show similar patterns outside social media. PocketOS, a software provider for car-rental businesses, said an AI coding agent, a tool that can write and run code like an assistant, ran one command that deleted the company’s main database and backups in seconds. The company said it had to piece records back together using an older off-site backup and other services like Stripe, plus calendars and emails.

Ars Technica also documented cases where AI coding helpers deleted a user’s database while trying to “fix” problems, and the user then depended on built-in rollback tools (like an undo button for a database) to restore what was lost.

What to watch

Expect more attention on basic safeguards, like stronger confirmation steps before deletions and clearer paths to reach a real person. For everyday users, the key question is whether platforms will offer faster, transparent ways to appeal mistakes, and whether exports and backups become easier to use.

Source: NYTimes