- Northflank told customers to assume all stored environment variables were decrypted. Users then reported that others were using their AI API keys.
- Northflank has not said how attackers got in or how many customers were affected, according to BleepingComputer.
- Leaders should ask who holds their AI keys, how fast they can be replaced, and who watches the AI provider's billing.
A key is a card that needs no PIN
An AI API key works like a payment card with no PIN. Whoever holds it can usually run up charges on the owner's account. That is why a problem at a hosting platform matters well beyond the platform itself.
Northflank is a cloud platform where developers run applications, containers, databases and AI applications. It disclosed a security incident on October 30 and told customers to replace their secrets. Later that day it said customers were reporting outside use of AI keys stored on the platform.
What Northflank has said
The timeline comes from Northflank's status page. Wiz, a cloud security company, sent a threat alert at 11:43 UTC on October 30. Northflank engineers reviewed it a minute later. A first network policy patch followed at 12:02, with more patching at 12:50 and 13:05. Northflank also asked Wiz's incident response team for help.
The advice grew firmer through the day. At 1:40 PM UTC the company told customers to rotate environment variables. At 1:58 PM it wrote: "Assume all environment variables have been decrypted." At 3:12 PM it said multiple users were confirming that their AI keys had been used.
Later updates added tools. Northflank enabled secret rotation for Postgres, MongoDB, MySQL and Redis. Every customer, free accounts included, now gets audit logs. These are activity records that help trace suspicious behaviour. The company also advised customers to turn off public access to add-ons.
What is still unknown
BleepingComputer reports that Northflank has not said how attackers got in. It has not said how many customers were affected. It has not said whether credentials other than AI keys were stolen or abused. BleepingComputer says Northflank did not reply to its questions.
On October 30, Northflank said it had no further confirmed findings. Read that as an open investigation, not a finished account.
Why environment variables matter
Environment variables are settings handed to an application when it starts. BleepingComputer notes that applications often keep database passwords, cloud credentials, tokens and API keys in them.
This is the publication's reading, not something the sources state. A platform that passes these values to running applications must be able to read them. That design means one compromise can expose many customers' secrets. Northflank has not said how decryption happened.
Wiz published a report called "Black October for AI Developer Platforms". It counts at least nine incidents at AI development platforms in October. They include Northflank, Railway, OpenCode, Warp and Lightning AI. The details differ.
BleepingComputer says Railway's case may also have exposed customer variables in readable form. OpenCode disclosed exposed AI API keys. Warp reported exposed AI agent logs and user telemetry.
Wiz says attackers in some of these cases went after variables holding AI service credentials. It ties this to LLMJacking. That means using stolen credentials to run AI models on another organisation's account. Northflank said its reports may fit that pattern. It also said similar IP addresses appear in the Wiz report.
You rent the platform, but the keys are yours to protect
This is the publication's view, not a finding from the sources. A platform can repair its own systems. It cannot undo use of a key that has already left. Charges for misused keys can show up on the customer's own AI provider account. Nothing in the sources says who ends up paying.
That is why Northflank tells customers to check their AI provider's usage and billing logs. The person who feels it first is probably not an executive. It may be the engineer who spots odd usage in a billing log. Or it may be the finance team asking why a model account shows charges nobody planned.
Questions to put to your team this week
Northflank's own guidance makes a sound checklist. Cancel old credentials with the company that issued them. Create replacements and update the secrets and workloads. Next, check what your AI provider's usage and billing records show, and look for activity you do not recognise. Report anything odd to both Northflank and the provider.
Beyond this one incident, ask four things.
First, which AI keys sit on third-party platforms, and who owns each one? Second, can we replace every one of them within a day? Third, does anyone watch AI billing for sudden jumps? Fourth, can we cap what any key is allowed to spend or reach?
If the answers are unclear, that is the finding. A secret you cannot quickly replace is a risk you carry on someone else's platform.
Produced by the WebPulse Newsroom with AI assistance from the original reporting credited below, and checked against that source by our editorial review. How we use AI.
Original reporting: Northflank.
