Northflank says rotate every stored secret now—what should customers do first?
Northflank says customers should assume user-defined environment variables were exposed. Reports of unauthorized AI API-key use make this an active remediation incident, but the company has not published how many customers or credentials were affected.
The short version
- What happened
- Northflank's status page remains in an Identified/Major outage state and tells customers to assume all user-defined environment variables were decrypted.
- Who it affects
- Developers and small teams may discover unauthorized AI charges before they notice broader account access. A rushed rotation can also cause outages, so incident owners need an inventory, clean administrative sessions and a record of which credentials were revoked, replaced and tested.
- What remains unproven
- Northflank has not published an affected-customer, key, workload, charge or data-access denominator.
Keep in mind
Northflank has not published an affected-customer, key, workload, charge or data-access denominator.
The root cause, dwell time and exact data-access path remain undisclosed. The instruction to assume exposure is broader than the currently proven downstream misuse.
The status page is changing. Readers should follow the maintained incident record rather than relying only on this snapshot.
Multi-source analysis · 2 linked sources. Primary analysed source date: 30 October 2026. Latest linked supporting source: 31 October 2026.
Keep following the evidence
Join The Impact Brief founding list for selected AI reports, their effect on people and the evidence limits. Read a sample first.
Read the full report · 4 min The evidence, methods, implications and limitations
The urgent task is revocation, not simply editing a platform field
Northflank is telling customers to treat every user-defined secret stored on its platform as compromised. Its public incident page says users have reported unauthorized use of AI API keys, while the investigation remains open and seven platform components are marked as experiencing a major outage. The current instruction covers environment variables in workloads, secret groups and global secrets. It also extends beyond AI providers: database passwords, cloud credentials, signing keys, webhooks and service tokens may occupy the same stores.
The first recovery step is to revoke each credential with the organization that issued it. Deleting or replacing a value only inside Northflank does not invalidate a copy that an attacker may already hold. Teams should generate a new credential from a clean, trusted session, update the relevant workload or managed add-on, and restart or redeploy only after the old value is unusable. Northflank has progressively enabled rotation for Postgres, MongoDB, MySQL, Redis, Minio and RabbitMQ add-ons, and it has enabled audit logs for free as well as paid customers.[1][2]
AI-key abuse is confirmed; the size and path of the breach are not
After revocation, customers need to review AI-provider usage and billing logs, cloud audit trails, database access, deployment history and outbound network activity for the period before rotation. Unexpected model calls can create direct charges, but a stolen key may also expose files, fine-tuning data or administrative functions depending on the provider and the key's permissions. Publicly exposed add-ons should be disabled during investigation. Organizations that find misuse should preserve logs and notify Northflank and the affected provider before normal retention windows erase evidence.
Northflank has published a response timeline beginning with a Wiz alert at 11:43 UTC on 30 October and several network-policy patches over the next 82 minutes. It has not yet published the vulnerability, dwell time, data-access path or final scope. Nor has it said whether every stored environment variable was actually read; its instruction is a precautionary assumption for safe recovery. The Breaking label therefore reflects the live remediation need, not a claim that every customer suffered downstream access. It will expire at 21:00 GMT unless a verified update warrants renewal or earlier removal.[1][2]
The distinction that matters
A safe order for rotating exposed secrets
Invalidate the old credential at its source before trusting a replacement stored on the platform.
Revoke
Disable the credential with its issuing provider
A copied value can no longer be used
Replace
Create a least-privilege replacement from a trusted session
Update only the workloads and add-ons that need it
Investigate
Review audit, usage, billing and network records
Identify misuse, preserve evidence and notify providers
Continue the story
Read next
Continue exploring ai infrastructure: compare the finding and its evidence limits.
How did 155 AI responses expose another customer’s data? A patched cache boundary failed
New analysis of disclosures dated 26–27 October: Cloudflare says one Workers AI model returned stale cache content from other customers in 155 responses between 27 August and 9 September. The service was patched on 9 September; the reports do not identify 155 customers or disclose the sensitivity of every leaked fragment.
Read the next reportExplore the full source and evidence record
What this means for people
- Developers and small teams may discover unauthorized AI charges before they notice broader account access. A rushed rotation can also cause outages, so incident owners need an inventory, clean administrative sessions and a record of which credentials were revoked, replaced and tested.
Global context
AI application stacks concentrate high-value provider keys beside ordinary cloud and database credentials. The incident is a reminder that model access should use narrowly scoped keys, spending alerts, short rotation cycles and separate administrative boundaries rather than one long-lived secret shared across development and production.
What the evidence does not yet show
- Northflank has not published an affected-customer, key, workload, charge or data-access denominator.
- The root cause, dwell time and exact data-access path remain undisclosed. The instruction to assume exposure is broader than the currently proven downstream misuse.
- The status page is changing. Readers should follow the maintained incident record rather than relying only on this snapshot.
What to watch next
- A root-cause report
- Affected-customer and key counts
- Evidence of non-AI credential misuse
- Completion of managed add-on rotation
- Service restoration
- Any regulatory notification
Living evidence record
Impact record IAI-0XFZBSG
Evidence stage
Announced
Confidence
Supported
Reporting basis
Multi-source analysis
Independent or research support
Present
Record status
Monitoring
Last checked
1 November 2026
Source trail
2 direct sources across 2 source types.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Evidence trail
Sources used for this report
Links checked 1 November 2026
Research question, topics and regions
Topics: Cloud security · API keys · Secrets management · Incident response · AI infrastructure · Northflank
Regions: Europe · United Kingdom · Global
This report is labelled multi-source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.