Security Guides

#secrets management #incident response #credentials #cloud #AI security

Northflank Security Incident (October 2026): Stored Environment Variables Treated as Decrypted, AI API Key Misuse Reported

Northflank says to assume all stored environment variables were decrypted, and users report AI API keys were used without permission. What to rotate and check.

Published 2026-11-01 Updated 2026-11-01 Last verified 2026-11-01 9 min read

On this page

For: developers and teams that deploy apps, jobs or databases on Northflank, especially those who keep AI service API keys, cloud credentials or database passwords in environment variables or secret groups. This article is based on Northflank's status page and press coverage, and does not cover attack techniques.

Bottom line

treat every secret you stored on Northflank as known to someone else, revoke it at the issuer, then check AI billing

According to Northflank, all environment variables should be assumed decrypted, and users have reported unauthorized use of AI API keys stored on the platform. The company has not said how many customers are affected or how the attacker got in.

Revoke each API key, cloud key and database password at the service that issued it, then set the new value on Northflank and redeploy
Cover workload variables, secret groups and global secrets, not only the ones you think matter
Open each AI provider's usage and billing page and look at the days from October 30 onward
Set spend limits and alerts at the provider, and use narrowly scoped keys
Use the add-on secret rotation the company has released, and turn off public access to add-ons you do not need exposed

Developing: this article will be updated as the organisation publishes more

Who is affected: developers and teams that deploy apps, jobs or databases on Northflank, especially those who keep AI service API keys, cloud credentials or database passwords in environment variables or secret groups.
Confirmed so farSee the table ↓: Listed as possibly affected
Environment variables on the platform
Workload, secret group and global secrets
Database and other add-on credentials
Not yet confirmed: Number of affected customers
How the attacker got in
Which AI providers' keys were used and how many
2 more
Whether credentials other than AI API keys were abused
A full incident report from the company
What to do nowSee the steps ↓: Revoke every secret you stored on Northflank, at the service that issued it
Rotate add-on secrets and close public access
Check AI provider usage and billing since October 30
2 more
Set spend limits and use least-privilege keys
Review activity with the audit logs
Where to verify (source)All sources ↓: Northflank, status page, "Security incident" (October 30–31, 2026)
Last checked: As of November 1, 2026

Follow-up plans and what we checked

On November 1, 2026 the status page still listed the incident as "Identified". Its latest entries are the release of secret rotation for add-ons one type at a time. The company has not published a full report, and a request for comment by one outlet had no reply when that outlet published on October 31.

What Northflank users should do today

Revoke every secret you stored on Northflank, at the service that issued it

The company says to assume all environment variables were decrypted, and to rotate secrets in workloads, secret groups and global secrets. Do it in this order: revoke the old key at the issuing service (AI provider, cloud, payment, email), create a new one, set it on Northflank and redeploy.

Editing or deleting the variable on Northflank does not stop a copy someone already has.

Rotate add-on secrets and close public access

The company released secret rotation per add-on type: PostgreSQL and MongoDB on October 30, MySQL and Redis on October 31, then Minio and RabbitMQ later on October 31. It also advised disabling public exposure of add-ons. Rotating a secret in the platform does not undo data that someone already read with the old one, so check the data and the accounts inside your databases too.

Check AI provider usage and billing since October 30

Look at each provider's usage, key list and billing page for requests you do not recognise, unusual models or regions, and a sudden rise in cost. The company's advice is to review billing logs and report suspicious usage to the provider and to Northflank. For why stolen AI keys turn into bills quickly, see stolen API keys and account bans.

Set spend limits and use least-privilege keys

At each AI provider, set a monthly budget cap and an alert, give each key a single purpose and the narrowest permissions it allows, and use separate keys per app and per environment. A smaller key limits what a leak costs. For the basics of keeping keys out of code and logs, see what's actually dangerous about .env and API keys.

Review activity with the audit logs

Since 01:04 UTC on October 31 the company has audit logs enabled for all customers, including free users. Open them and look for changes you did not make: new members, tokens, secrets, domains or deployments.

What happened (from Northflank's status page)

The following is based on the status page entry titled "Security incident". Times are UTC. The entries from 11:43 to 13:05 show no date on the page; they appear inside the timeline of the company's October 30 13:40 UTC entry, and press coverage places them on October 30.

October 30, 11:43
A threat alert from a security vendor is raised.
October 30, 11:44
The company's engineers start reviewing the alert.
October 30, 12:02
An initial network policy patch is deployed. Further patches follow at 12:50 and 13:05.
October 30, 13:40
Customers are told to rotate environment variables immediately.
October 30, 13:58
Status changes to "Identified". Customers are told to assume all environment variables were decrypted and to disable or rotate all keys.
October 30, 15:12
Secret rotation is enabled for PostgreSQL and MongoDB. Multiple users report their AI keys were used.
October 30, 23:38
The company says users reported unauthorized use of AI API keys stored on the platform, and advises rotating and revoking credentials and reviewing provider billing logs.
October 31, 01:04
Audit logs are enabled for all customers, including free users.
October 31
Secret rotation for MySQL and Redis (13:34), Minio (22:32) and RabbitMQ (23:31) is released to all customers.

Assume decrypted

Environment variables in workloads, secret groups and global secrets

Abuse reported by users

AI API keys stored on the platform; providers and counts not disclosed

Either way, revoke at the issuer first, then replace.

What the company says to assume, and what users reported

What the status page says (Northflank, October 30–31, 2026)

What happened: A security incident; the company says to assume all environment variables were decrypted
Abuse reported: Users reported unauthorized use of AI API keys stored on the platform
Scope of guidance: Environment variables in workloads, secret groups and global secrets
Response: Network policy patches, customer notice to rotate, add-on secret rotation, audit logs for all customers
Not disclosed: Number of affected customers, how the attacker got in, which AI providers were involved

What is known and what is not

The state as of November 1, 2026. Items marked "not stated" are not in the status page entries we checked; they are not confirmed as safe.

Item	What the company says	Status
Environment variables on the platform	Assume all were decrypted	Possible (company)
Workload, secret group and global secrets	Rotate all	Possible (company)
AI API keys stored on the platform	Reported by users ("multiple users are confirming their AI keys have been used"), as posted by the company	Reported by users (posted by company)
Database and other add-on credentials	Rotation released per add-on type; public exposure should be disabled	Possible (company)
Credentials other than AI keys	Not stated whether they were abused	Not stated
Number of affected customers	Not disclosed	Not stated
How the attacker got in	Not disclosed	Not stated
AI providers involved	Not named	Not stated

How to read this: 'possible' still means rotate

The company's own guidance is to assume the variables were decrypted. Waiting for proof of misuse for a particular key does not make sense, because an unused leaked key can be used later.

What leaked environment variables can lead to

In general, secrets taken from environment variables can be misused as follows. This does not mean all of it has happened here.

AI service API keys: someone else sends requests on your account and you receive the bill, or the key is resold
Database passwords and connection strings: reading or changing your app's data, including your own users' records
Cloud and payment keys: creating resources or making calls that cost money
Signing and encryption keys: forging session tokens or decrypting data your app protects

A security vendor reported multiple incidents at AI developer platforms in October (per press coverage). The company says the reports may be consistent with a broader pattern of attacks on AI credentials described by a security vendor; the vendor has not confirmed a single actor across all cases (press). For other cases this month, see Zeabur and Railway.

For operators: lessons from what the company disclosed

These points come from the facts the company published. They are not a further judgement of Northflank.

A vendor's alert is what started the response. Route alerts from your cloud security tooling to people who can act within minutes; the first patch here followed the alert by 19 minutes.
Offer secret rotation for every kind of stored credential before you need it. Rotation for add-ons was released one type at a time across two calendar days (October 30 15:12 to October 31 23:31 UTC). Customers need a way to rotate each secret type quickly.
Turn audit logs on for everyone. The company enabled them for all customers, free included, only after the incident began.
Treat stored secrets as the first thing an intruder tries to use. Users reported misuse of AI API keys within hours of the first alert.

Sources (public records)

The facts in this article are based on the public information below. Intrusion routes that the company has not confirmed are not speculated about.

Northflank, status page, "Security incident" (October 30–31, 2026) — status.northflank.com
BleepingComputer, "Northflank security breach leads to abuse of customers' AI API keys" (October 31, 2026) — bleepingcomputer.com

Update history

2026-11-01: First version. Based on Northflank's status page as of November 1, 2026 and one press report. This article will be updated when the company publishes more.

Latest breaking reports

All breaking reports →

FAQ

QWhat happened at Northflank?

According to Northflank's status page, a security vendor's threat alert arrived at 11:43 UTC on October 30, 2026. The company patched its network policy from 12:02 UTC, told customers at 13:40 UTC to rotate environment variables immediately, and at 13:58 UTC told them to assume all environment variables were decrypted. The number of affected customers and the way in have not been disclosed.

QWhat should I rotate?

The company says to treat environment variables for workloads, secret groups and global secrets as compromised and to rotate all secrets. Revoke each key at the service that issued it first, then set the new value and redeploy. For databases and other add-ons, the company has released secret rotation per add-on type, and it advises turning off public exposure of add-ons.

QWere my AI API keys used?

The company's 15:12 UTC entry already says multiple users were confirming their AI keys had been used, and its 23:38 UTC entry says users reported unauthorized use of AI API keys stored on the platform. It has not said which providers or how many keys. Check each AI provider's usage and billing logs for requests you do not recognise, revoke the key, and report suspicious activity to the provider and to Northflank.

QHow was Northflank breached?

Not disclosed. As of November 1, 2026, the status page and the press coverage do not say how the attacker got in, how long they had access, or how many customers were affected.

QDo I need to do anything if I have no AI keys?

Yes. The company's guidance covers all environment variables, not only AI keys. Database passwords, cloud credentials, tokens and signing keys stored in variables should be rotated too. It enabled audit logs for all customers including free users, so you can review activity from the account side.

QWhat other steps limit the damage of a leaked AI key?

Set a monthly spend limit and alerts at the AI provider, give each key the narrowest permissions and a single purpose, use separate keys per app, and watch the billing page for the days since October 30. A leaked key stays valid until you revoke it at the provider.