Skip to content
Submit incident
AII-1001-15359
Well documentedThree or more independent publishers report the core facts.Supporting sources are pending verification.

Claude AI accidentally deletes production database with 2.5 years of user data

Claude Code, an AI coding agent from Anthropic, deleted the production database of DataTalks.Club after the founder delegated infrastructure commands to it without adequate safeguards. The incident resulted in the loss of 2.5 years of user data, though the founder was able to restore it using AWS recovery tools.

6 March 2026 Β· Anthropic Β· Reliability and hallucination
6 March 2026
AII-1001-15359
Quick facts
Category
Reliability and hallucination
Country
Germany
Date
6 March 2026
Drafted by
Raidu Incident Detection

What happened

Alexey Grigorev, founder of DataTalks.Club, delegated Terraform infrastructure-as-code commands to Claude Code, including plan, apply, and destroy operations. Grigorev instructed the AI in a way that led it to execute a "destroy" command, wiping out the entire production database.

Reported impact

The incident erased 2.5 years of user data from the learning platform, including homework submissions, projects, leaderboards, and automated backups. Grigorev was able to restore the database using AWS recovery tools.

Index analysis

Grigorev admitted the root cause was removing the final safety layer by allowing the AI to run commands directly without manual oversight, bypassing Terraform's built-in safeguards that preview changes before applying them.

Governance controls

Grigorev shared the incident on LinkedIn and in a blog post, noting the exposure of serious gaps in the system: untested backups, insufficient protections against deletion, and how literally the AI follows instructions.

Classification

Category
Reliability and hallucination
Industry
IT infrastructure and hosting
Country
Germany
Harm status
Actual harm
Intent
Unintentional
Timing
Post-deployment
Harm types
Economic or property
Harmed
Business
Failure modes
Agent misaction, Overreliance, Unsafe autonomous behaviour
AI systems
Claude Code

Sources and evidence

Sources not yet published

This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.

How sources are verifiedSuggest a source for this incident