If you run anything on Northflank, revoke and replace every secret stored there, starting with AI provider keys, and check your AI billing for usage you cannot explain. That is the advice the cloud deployment platform has given all customers after the Northflank security incident, which it disclosed on Friday. “Users have reported unauthorised use of AI API keys supplied to the Northflank platform,” the company said in an update on its status page at 23:38 GMT that night.
Northflank says its guidance “remains unchanged”: “assume all environment variables have been compromised and rotate all secrets and credentials immediately”. What it has still not given is an explanation. There is no root cause, no named attacker and no count of affected customers. The incident remained open, rated a major outage, when TSN checked the page at 18:55 GMT on Saturday.
In brief
- Northflank says all customers should assume their environment variables were decrypted, and users report that AI API keys stored on the platform have been misused.
- Rotating a secret only helps once the old one is revoked with the provider that issued it; Northflank has added audit logs and rotation tools to help.
- Northflank has disclosed no root cause, attacker or customer count. Wiz, whose threat alert started Northflank’s response, lists the incident among nine at AI developer platforms this month.
What Northflank customers should do now
Northflank, whose registered office is in London, lets developers build and run applications and databases. Like other platforms of its kind, it stores the environment variables those applications need, such as database passwords and API keys for outside services. Its updates add up to five steps.
- Revoke AI keys at the source first. Northflank tells customers to “Revoke existing credentials with their issuing providers, generate replacements, and update the affected secrets and workloads in Northflank.” Changing a value only on Northflank leaves the old key working for whoever holds it.
- Check AI usage and billing. It asks customers to “Review your AI provider usage and billing logs for unexpected activity”, and to report anything suspicious to Northflank support and to the provider concerned.
- Rotate everything else. That covers secrets in workloads, secret groups and global secrets: database passwords, cloud credentials, code host tokens and signing secrets. Northflank enabled secret rotation for Postgres and MongoDB for all users on Friday, and since 13:34 GMT on Saturday, “Secret rotation for MySQL and Redis is now available to all customers.”
- Close public access. Northflank says to “Disable public access to addons where enabled.”
- Read the audit logs. At 01:04 GMT on Saturday, Northflank said: “We have enabled audit logs for all customers, including free users.” Check them, and the logs of the services your keys reach, for logins, deployments or data access you cannot account for.
How the Northflank security incident unfolded
The first public notice, at 13:40 GMT on Friday 30 October, said only: “Rotate environment variables immediately.” Beneath it, Northflank posted a timeline. It received a “Wiz threat alert” at 11:43 GMT, from the cloud security company, and engineers reviewed it a minute later. An “Initial network policy patch” followed at 12:02, more patching at 12:50 and 13:05, and at 13:00 Northflank requested “Wiz CIRT support”, meaning Wiz’s incident response team.
At 13:58 the wording hardened: “Assume all environment variables have been decrypted. Disable/rotate all keys.” By 15:12 Northflank was relaying that “Multiple users are confirming that their AI keys have been used.” At 19:28 it said: “Our investigation is ongoing. We have no further confirmed findings to share at this time.”
The 23:38 update turned those user reports into a written warning about AI keys. It added that “These reports may be consistent with the broader pattern of attacks targeting AI credentials”, and linked to a report Wiz published the same day. All seven Northflank Platform components, including Networking, Services, Addons, Jobs and Builds, were still listed as in major outage on Saturday, while the Northflank app and API were shown as operational.
Wiz counts nine AI developer platform incidents in October
In that report, Wiz researchers Benjamin Read and Rami McCarthy wrote that “At least nine AI Developer Platforms have disclosed incidents over the course of the month”. Their list runs from Fireworks AI on 6 October to “Northflank (Oct 30)”. They wrote that “All public reports highlight targeted exploitation, explicitly hunting stored environment variables, with a focus on those granting AI access”, while noting “there is no single vulnerability or dependency linking these attacks”.
Wiz does not say what happened inside Northflank, and Northflank’s own wording is careful: the user reports “may be consistent” with that pattern. Neither company has published a finding that ties this incident to the others.
A month of breaches has made AI keys the first thing to revoke
The Northflank security incident follows a run that TSN has covered this autumn. Railway said an attacker may have decrypted some customer environment variables, Zeabur said an attacker had read platform environment variables, and Fireworks AI said an intruder captured customer API keys stored in evaluation jobs. In each case, the credentials customers had entrusted to the platform were the prize.
AI keys head the queue because they turn into money fastest. They are billed by use, so anyone holding one can spend the owner’s budget at once, with no further break-in needed. A stolen database password is serious, but exploiting it takes work; a stolen model key is simply spent, and the first sign is often a bill.
That makes the fix as much about records as technology. Rotation on Northflank closes nothing until the old credential is revoked where it was issued, so teams need to know every outside service their apps call and who owns each key. For compliance teams, a platform breach like this is also a third-party risk test: whether the supplier register lists which secrets sit with which hosting provider, and whether anyone can show the old keys were actually killed.
What Northflank has not said
Northflank has not given a root cause or said how environment variables came to be at risk. It has not named an attacker, said how many customers are affected, or said whether anything beyond secrets was accessed. It has not published what Wiz found. BleepingComputer reported on Saturday that Northflank had not answered its questions about the breach and its impact on customers.
As with Allium’s unauthorised access incident, the first notices give customers instructions before explanations. TSN tracks incidents like this on its ransomware and cyber attacks hub. The next thing to watch is whether Northflank moves the incident past “Identified” with a written account of how the attackers got in.
Sources
- Northflank status page, “Security incident” (updates 30 October 2026 13:40 GMT to 31 October 2026 13:34 GMT; checked 31 October 18:55 GMT)
- Wiz, Benjamin Read and Rami McCarthy, “‘Black October’ for AI Developer Platforms” (30 October 2026)
- BleepingComputer, Lawrence Abrams, “Northflank security breach leads to abuse of customers’ AI API keys” (31 October 2026)
- Northflank privacy policy (registered office)
Related stories
- Fireworks AI Says Intruder Captured Customer API Keys From Eval Jobs
- Railway Says Attacker May Have Decrypted Some Customer Environment Variables
- Allium Revokes API Keys After Unauthorised Access to Its Systems
- Wiz Counts Nine AI Developer Platform Breaches in ‘Black October’
Follow @tsnmediahub on X for breaking AI news as it happens.
Follow on X
