Railway, a cloud platform that hosts applications for developers, is urging all of its users to replace the passwords and keys they store with it after an intruder may have reached and partly decrypted them, BleepingComputer reported Thursday. Railway says the break-in is one of a series of attacks on platform companies aimed at AI inference providers, the services that run AI models for paying customers. Any organization with software on Railway should assume its stored credentials are exposed until told otherwise.
What Happened
In a security disclosure dated Oct. 21, Railway said it detected unusual activity that morning on systems that store customer environment variables. These are settings an application reads when it runs, and they often hold secrets such as API keys, the codes that let software use a paid service, along with database and container registry credentials.
Railway said an unauthorized third party may have accessed the encrypted variables and, for some of them, obtained the decrypted contents. Its initial analysis suggests a subset of users is affected. It has not said how many, and BleepingComputer's figure of about 4.8 million users and more than 16 million hosted services describes the whole platform, not the number affected.
The company told all users to rotate every stored variable, including private registry credentials, and to review their service logs for unusual activity. It said it will email customers once it confirms which accounts were involved.
Railway paused all deployments during the investigation. By 08:44 UTC Oct. 22, builds had resumed for paying Pro and Enterprise workspaces while deployments stayed delayed, BleepingComputer reported, with Hobby, Free and Trial workspaces to follow in that order.
The disclosure does not explain how the intruder got in, and it does not mention law enforcement.
The AI Connection
Railway said it believes the attack is "part of a series of attacks recently targeting platform companies" with the aim of compromising AI inference providers. It has notified those upstream providers so they can revoke all keys as a precaution. Railway did not name the providers or the other platforms.
The pattern matches other recent incidents. The Primary Wire reported Oct. 19, citing BleepingComputer and TechCrunch, that attackers took AI provider keys stored on the OpenCode Console platform and that affected users saw unauthorized charges within hours. Two days later, the publication reported that a hijacked release of the AI gateway LiteLLM carried credential-stealing malware.
Why It Matters
A hosting platform holds the secrets that its customers' applications use to reach databases, code registries and paid AI services. One intrusion there can open doors into many companies at once, and stolen AI keys can quickly turn into bills or resold access.
For executives, the practical issue is speed and inventory. Rotating secrets means knowing every key stored on Railway, who owns it and which downstream systems depend on it. Companies without that list will take longer to close the exposure, and may miss some keys entirely. The incident is also a reason to check whether AI provider keys carry spending caps and alerts.
What to Watch
- Railway's emails to affected customers and any count of accounts involved.
- An explanation of how the intruder got in and whether the decryption keys themselves were exposed.
- Whether other platforms or AI providers confirm they were part of the same campaign.
