Skip to content
NewCybercrime Report 2026 · Eight Chapters · A Post Is Not An Incident → Explore The ChaptersCybercrime Report 2026 → Explore

Railway Cloud Platform Security Incident (October 2026)

On 21 October 2026 the cloud platform Railway said an unauthorized third party may have accessed encrypted customer environment variables and, for some, obtained the decrypted contents.

Published
Updated
Reading Time
6 min read
Dark glass rail track with a cracked junction with red neon lines, cover art for: Railway security incident (2026)

Railway asked all users to roll every variable, API key and registry credential, paused deployments, and reset bucket credentials, GitHub tokens and API tokens itself. In a preliminary report of 22 October it said at most 1.85% of workspaces, about 100,000, had environment variables potentially compromised.

What happened

Railway says that on 21 October 2026 it disclosed anomalous activity involving systems that store customer environment variables. Its preliminary report of 22 October says it became aware of the attack on 20 October at 19:53 UTC. Its investigation to date indicates an unauthorized third party may have accessed encrypted environment variables and, for some, obtained the decrypted contents.

Railway says environment variables may contain secrets such as API keys, and database and registry credentials. It believes the impact may be limited to a subset of users, but it recommends that all users act as a precaution.

Railway paused all deployments while it investigated. On 22 October its status page showed builds returning in stages. Railway published a preliminary report the same day and says a final forensic report will follow. Wiz counts Railway among at least nine AI developer platforms with incidents in October 2026; Northflank is another, covered in our report on the Northflank incident.

For the order in which to change your secrets, see our Railway breach rotation checklist.

Timeline

  1. 21 Oct 2026, 07:24 UTC

    Railway's status page opens a platform outage entry. Later entries say the dashboard, API and CLI are unavailable and builds are paused. Railway does not say in the pages we read whether this is linked to the anomalous activity.

  2. 21 Oct 2026, 20:19 UTC

    Railway says it is invalidating OAuth sessions because of impact from adjacent cache systems. Affected users are signed out.

  3. 21 Oct 2026, 21:59 UTC

    Railway publishes its security notice and recommends revoking all environment variables, API keys and private registry credentials stored in Railway.

  4. 21 Oct 2026, 22:10 UTC

    Railway says it is rotating all bucket credentials. At 22:45 UTC it adds that customers will see the revocations by a Railway employee in audit logs.

  5. 22 Oct 2026, 01:22 UTC

    Railway publishes security guides, among them PostgreSQL credential rotation and checks for unexpected usage of AI provider keys.

  6. 22 Oct 2026, 03:32 UTC

    Builds for Enterprise workspaces become available again.

  7. 22 Oct 2026, 04:17 UTC

    Railway invalidates all API tokens. Users must create new ones.

  8. 22 Oct 2026, 17:43 UTC

    Railway says the API is fully restored. API tokens and Project Tokens created before the incident remain invalidated.

  9. 22 Oct 2026, 21:59 UTC

    Builds are enabled for Hobby, Pro and Enterprise workspaces. Free and Trial workspaces are to follow, and new signups remain paused.

How the attack worked

Railway’s notice of 21 October says only that the activity involved systems that store customer environment variables. It adds that the third party may have obtained the decrypted contents of some of them. Its preliminary report of 22 October adds that it first saw an attempted privilege escalation in its build infrastructure on 20 October. Further routes of access were identified on 21 October.

It says the attacker pulled encrypted secrets, then reached CI credentials and decrypted artifacts that held more API access. A database query of encrypted environment variables followed, and a subset was decrypted using the compromised credentials. Railway says it cannot yet share the specific details of the exploits.

Railway says it believes the incident is part of a series of attacks on platform companies, aimed at compromising AI inference providers. It notified those upstream providers so they could revoke keys early.

Who says who was behind it

No attacker is named. Railway states a belief that this is part of a series of attacks on platform companies, aimed at AI inference providers, but it does not name an actor. Its 22 October preliminary report also names none. It describes an external hostile party that Railway believes had access to "abliterated models and substantial compute resources" and was likely looking for AI and platform API keys.

Impact

Railway says the impact may be limited to a subset of users and that it will email customers it confirms as affected. Its preliminary report says at most 1.85% of all workspaces, about 100,000, had some environment variables potentially compromised. Vendors received reset requests for up to all stored keys, so more users than that heard from vendors.

Railway also says it is contacting customers it rates as high risk, a group larger than that 1.85%. It dates the attack window from 20 October 13:58 UTC to 22 October 03:32 UTC and has not published the number of variables decrypted. Customers lost the use of deployments while builds were paused.

Government and regulator response

Railway’s own response, as of 28 October 2026:

  • Railway paused all deployments on 21 October and re-enabled builds in stages from 22 October (status page).
  • It rotated every storage bucket credential and reset the GitHub tokens it held for customer accounts (Railway guide, 22 October).
  • It invalidated all API tokens and Project Tokens created before the incident (status page).
  • It published step-by-step guides for PostgreSQL, MongoDB, Redis and MySQL credentials, and for AWS, Google Cloud, GitHub, Stripe, OpenAI, Anthropic, Grok and OpenRouter.
  • It rolled all keys, then rolled them again after confirming the intruder was out, and rebuilt its infrastructure tooling from validated source (preliminary report, 22 October).
  • It moved builders from multi-tenant infrastructure to single-tenant environments ahead of plan, and added a way to revoke Railway application permissions in bulk (preliminary report).
  • It engaged an external forensics firm. On 26 October it said services were fully restored for existing users, with new signups still paused (report addendum; status page, 27 October).

Lessons

  1. A platform that holds your secrets is a supplier

    Railway stored environment variables for its customers, so one incident put every customer's keys in question. Track such platforms as third-party risk.

  2. Rotation is only done after a redeploy

    Railway's guides say containers read variables only at startup. A service keeps using the old secret until it restarts.

  3. Rotate and then look back

    A new password stops future access but does not show whether the old one was used. Railway's guides pair each rotation with a log review at the provider.

  4. Use references instead of pasted values

    Railway tells users to replace pasted connection strings with reference variables, so the next rotation reaches every service automatically.

What to check in your own organization

  1. List every secret stored in Railway

    Open each service's Variables tab and the shared variables in Project Settings. Note pasted connection strings and copies outside Railway, such as CI jobs and BI tools using the public database URL.

  2. Rotate databases with Regenerate

    In the Database tab, use Regenerate rather than editing the password variable or running ALTER USER. Redeploy every service that references it.

  3. Rotate cloud, payment and AI keys at the provider

    Deactivate the old AWS key, disable the old Google Cloud key with the gcloud CLI, rotate the Stripe key, and delete and recreate AI provider keys.

  4. Review provider logs for the exposure window

    AWS CloudTrail event history by key ID in every Region, Google Cloud audit logs by service account, the GitHub security log (90 days) and organization audit log (180 days), and Stripe request logs for the old key.

  5. Check sessions on the database

    After a PostgreSQL rotation, list pg_stat_activity. A session you do not recognize connected with the new password.

  6. Create new Railway tokens and reconnect GitHub

    All earlier API tokens are invalid. Reconnect GitHub in Account Settings and replace any ghcr.io token with a new one that has the read:packages scope.

Frequently Asked Questions

What did Railway say was accessed?

Railway says an unauthorized third party may have accessed encrypted environment variables and, for some of them, obtained the decrypted contents.

Who is behind the Railway incident?

No one has been named. Railway believes it is part of a series of attacks on platform companies aimed at AI inference providers.

Do all Railway users need to rotate secrets?

Railway recommends it for all users as a precaution, and says it will email customers it confirms as affected.

Is Railway working again?

At 21:59 UTC on 22 October Railway's status page said builds were enabled for Hobby, Pro and Enterprise workspaces. Free and Trial were still to follow.

Has Railway published an incident report?

Railway published a preliminary report on 22 October and added an update on 26 October. It says its forensic investigation is still ongoing and that a follow-up will come when it concludes.

Sources

  1. Security Disclosure: Wednesday October 21stRailway Blog
  2. Railway Platform OutageRailway Status
  3. Redeploy Buckets and Reconnect GitHub After the Credential ResetRailway Blog
  4. Rotate PostgreSQL Database CredentialsRailway Blog
  5. Check for Anomalous AWS AccessRailway Blog
  6. Check for Anomalous GCP AccessRailway Blog
  7. Check for Anomalous GitHub AccessRailway Blog
  8. Check for Unexpected Usage in StripeRailway Blog
ShareLinkedInX
Brandefense logo

Stay Ahead of Digital Threats.

Take control of your digital security with an exclusive demo of our AI-driven threat intelligence platform.

Open The Full Demo Form →