TechBroz · techbroz.com47 incidents on record · updated as they happen
← Today in agent chaos

Agent chaosVerified·1 Nov 2026

Cloud platform Northflank told every customer to assume their secrets were decrypted, as stolen AI API keys started getting used

Cloud platform Northflank told every customer to assume their environment variables had been decrypted after a breach on 30 October, and users then reported their stored AI API keys being used by someone else.

Northflank is a cloud platform where developers deploy apps, databases and background jobs, and where they store the secrets those apps need, including API keys for AI models. On 30 October 2026 it posted a security incident that ended with the worst instruction a hosting company can give: rotate everything.

What happened

Northflank says a threat alert from the cloud security firm Wiz arrived at 11:43 UTC, and its engineers looked at it a minute later. It pushed a first network policy patch at 12:02, more patches at 12:50 and 13:05, and asked Wiz's incident response team for help at 13:00. At 13:40 it told customers to rotate their environment variables immediately.

Eighteen minutes later the advice got blunter: "Assume all environment variables have been decrypted. Disable/rotate all keys." That covers secrets on workloads, secret groups and global secrets.

Where the AI keys come in

By the afternoon, Northflank said multiple users were confirming that their AI keys had been used. Late that night it repeated the warning in plain terms: users had reported unauthorised use of AI API keys supplied to the platform. It told everyone to revoke those keys with the provider that issued them, create new ones, and go through their AI usage and billing logs for anything unexpected.

Northflank hasn't said how the attacker got in, whether any database contents were read, or how many accounts were hit. It says its investigation is ongoing.

What Northflank did

It switched on one-click secret rotation for Postgres and MongoDB the same day, then for MySQL, Redis, MinIO and RabbitMQ add-ons by the end of 31 October. It turned on audit logs for every customer, including free users, and told people to turn off public access to database add-ons where they had enabled it.

It isn't the only one

Wiz, the firm that raised the alert, published research the same day saying at least nine AI developer platforms disclosed incidents during October, with Northflank the latest. It says all the public reports describe attackers hunting stored environment variables, especially ones that grant AI access, and that three of the companies reported signs of an "agent swarm" attack.

Wiz also traced what happens to keys after they're stolen. In two cases from one of the breached platforms (it doesn't say which), each key was first tested within 30 minutes of being taken. Days later, one made about 800,000 calls to Amazon's Bedrock AI service in 30 hours, and the other about 1.7 million in 13 hours. Wiz estimates those two keys alone ran up more than $100,000 at list price.

Our take

This is our opinion. An AI API key is now one of the most valuable secrets you own, because it turns straight into money for whoever holds it. Give each app its own key with a spending cap, set billing alerts that wake someone up, and know how to rotate every key you have in minutes, not days.

Lesson for builders

Treat AI API keys like cash: one key per app, hard spending caps, billing alerts, and a rotation plan you've actually tested.

Fail card: Cloud platform Northflank told every customer to assume their secrets were decrypted, as stolen AI API keys started getting used
The fail card

This story in one shareable card. Tap to open it full size.

Share on X All fail cards →

More agent chaos

Every day of agent chaos →

More fails

Get TechBroz Weekly free → All 47 fails