Skip to content
NewA calmer, faster ZeroHour
StoryData breachhighExploited in the wild

Railway says October breach may have exposed secrets and $740,000 in AI charges

4 sources · 4 articlesFirst reported Updated ()
What's new

November 11: The Hacker News says AWS and Google started fraud refunds, OpenAI reviews claims individually, and Railway will pay charges Anthropic and OpenRouter will not refund.

Merged summary · x-ai/grok-4.7 · rewritten as coverage arrivesSummaries by Vani AI

Railway says an October breach may have decrypted secrets in up to about 100,000 workspaces and driven about $740,000 in AI charges.

Railway said on November 10 that an attacker used a previously unknown build-isolation flaw, an over-trusted credential, and a faulty OpenID Connect trust setting to decrypt customer environment variables. The first successful exploit was at 07:28 UTC on October 20; decryption followed on October 21. Railway puts potentially affected workspaces at an upper bound of 1.85%, about 100,000, not confirmed thefts, and estimates about $740,000 in unauthorized AI charges. It reset linked tokens for 1,852 users, restored builds by October 22, and found no stolen-credential logins or repository or database downloads. GBHackers, citing an October 21 notice, says charges were unconfirmed and the outage was resolved October 23; BleepingComputer dates disclosure to October 22 and links the theft to Wiz's Black October campaign.

Key facts
  • First exploit at 07:28 UTC on October 20; decryption followed October 21.
  • Attacker read histories from 353 builds that included secrets.
  • Linked-account tokens for 1,852 users were reset, including GitHub OAuth and AI keys.
  • Railway found no stolen-credential logins and no customer database access.
  • AWS and Google started fraud refunds; OpenAI reviews claims case by case.
  • Customers must rotate exposed secrets; Railway will cover some unreimbursed charges.
Analyze this story with Vani AIImpact, who is affected and what defenders should do, in a new chat

Timeline

dated events in the reporting · oldest first
  1. Oct 20, 2026First successful exploit at 07:28 UTC · Cyber Security News
  2. Oct 21, 2026Faulty OIDC setting exposed a short-lived cloud credential · Cyber Security News
  3. Oct 21, 2026Attackers removed after the intrusion · BleepingComputer
  4. Oct 22, 2026Builds restored; disclosure dated this day · BleepingComputer
  5. Oct 23, 2026Outage marked resolved · GBHackers
  6. Nov 10, 2026Railway published its incident report · Cyber Security News

What each outlet adds

beyond the shared facts
  • Cyber Security NewsFirst exploit time, 353 build histories, and the 1.85% bound.
  • BleepingComputerAbout 26-hour intrusion, October 22 disclosure, Black October link.
  • The Hacker NewsRefund split among AWS, Google, OpenAI, Anthropic, and OpenRouter.
  • GBHackersOctober 21 notice left charges and scope unconfirmed; resolved October 23.
Vendors
Railway
Products
Railway
Victims
Railway
Organizations
AWSGoogleOpenAIAnthropicOpenRouterSpaceXSIWiz

Coverage

oldest first · one row per article · UTC
  1. · 1d ago Cyber Security NewsFirst report

    Railway says attacker stole secrets and incurred about $740,000 in charges

    An upper limit of about 100,000 workspaces may have had environment variables decrypted; Railway reset 1,852 tokens.

  2. · 22h ago BleepingComputer

    Railway breach leads to $740,000 in fraudulent AI charges

    Attackers stole environment secrets from at most about 100,000 workspaces after two infrastructure flaws.

  3. · 13h ago The Hacker News

    Railway says stolen customer AI keys drove $740,000 in unauthorized charges

    An attacker decrypted secrets in at most about 100,000 workspaces; Railway will cover some charges providers will not refund.

  4. · 12h ago GBHackers

    Railway breach may have exposed customer API keys and secrets

    After an October 21 disclosure, Railway paused builds, rotated platform credentials, and told customers to rotate secrets.