Railway, a platform that lets developers build and host apps in the cloud, disclosed in a blog post Wednesday that unidentified hackers “may have accessed encrypted environment variables and, for some of them, obtained the decrypted contents.” These environment variables typically contain passwords, API keys, and other secrets that apps use to connect to services like databases and cloud storage.
The company recommended that “all users” rotate their keys, warning that attackers could use the stolen secrets to log in and access data as if they were the rightful owner. Railway said its initial analysis suggests the breach was limited to a “subset of users,” but it did not provide a specific number or disclose the root cause.
Railway said the incident was likely part of a campaign targeting platforms “with the aim of compromising AI inference providers”, a technique known as AI distillation, where attackers trick AI systems into leaking sensitive data through malicious prompts.
The breach follows a similar incident this week at open-source AI coding tool OpenCode, which said hackers stole more than 1% of its customers’ API keys and other secrets. OpenCode attributed that attack to AI distillation. Earlier this week, the US Cybersecurity and Infrastructure Security Agency (CISA) warned that China-based AI companies are conducting large-scale attacks on US companies with the aim of improving their own models.
The Railway attack appears to be separate from two critical security vulnerabilities in the PHP programming language that were publicly disclosed this week. Patches for the bug, dubbed “PHP2Shell,” were released Wednesday, with security researchers at Amazon reporting evidence of exploitation four days earlier.
