Skip to page content

AEGLIX MIND

Connected intelligence.Security research, connected

AEGLIX news · Global

Railway breach exposes 100,000 workspaces and drives $740,000 in fraudulent AI charges

Based on reporting from BleepingComputer

Published by the source: Tue, 10 Nov 2026 17:20:19 -0500

AEGLIX News

On October 22, Railway revealed a security breach where threat actors stole customer authentication secrets and racked up roughly $740,000 in fraudulent AI consumption, putting credentials for as many as 100,000 workspaces at risk. The company also determined there was no indication the intruders downloaded client Git repositories or reached customer databases.

The intrusion started on October 20 when adversaries leveraged a zero-day flaw in Railway's build infrastructure to break filesystem isolation and seize an internal credential. With that credential, they accessed stored build histories and pulled Railway's internal source code, studying it to uncover further compromise paths. A second vulnerability, rooted in a misconfigured trust relationship, then yielded credentials capable of decrypting customer environment variables. Railway confirms attackers accessed decrypted environment variables for no more than 1.85% of workspaces, about 100,000, which represents the upper bound of potential impact rather than confirmed thefts.

The event turned into an approximately 26-hour confrontation between Railway engineers and threat actors who employed AI-driven agents to discover and exploit vulnerabilities throughout the infrastructure. The attackers hunted specifically for AI provider credentials, though some queries also targeted cloud, database, and payment service credentials. The purloined credentials fueled roughly $740,000 in unauthorized AI usage, and Railway will absorb any charges the providers refuse to refund. The breach also exposed linked-account tokens belonging to 1,852 users, including GitHub OAuth tokens and AI agent keys.

Railway says attackers were fully removed from its systems on October 21 and all exploited vulnerabilities have since been remediated. The company reset the exposed linked-account tokens belonging to 1,852 users and notified affected users about the compromise.

Report prepared 2026-11-10 · Based on reporting from BleepingComputer. Source publication date above; subsequent developments may not be reflected.

Sources

Read original source · BleepingComputer (opens in a new tab)

Privacy policy

Explore the story

Consult the cited publisher for the original reporting, updates and supporting evidence. Check the publication date and source before acting on a headline.

These are research tools; no CVE, indicator, or campaign relationship is inferred from this article.

More from BleepingComputer

← Back to security news