Northflank Breach: Stolen AI API Keys Abused, Rotate All Secrets

Cloud hosting platform Northflank has told every customer to assume their stored secrets were stolen after a Northflank breach that began on October 30, 2026. Several users have already reported that AI API keys they kept on the platform were used without permission, and the company says all environment variables should be treated as decrypted and rotated immediately.

Northflank lets developers deploy apps, containers, databases and AI workloads without running their own servers. Apps on such platforms usually keep passwords, database credentials and keys for services like OpenAI, Anthropic or Amazon Bedrock in environment variables, so a leak can turn straight into someone else’s bill.

How the Northflank Breach Unfolded

According to the timeline on the Northflank status page, cloud security firm Wiz sent a threat alert at 11:43 UTC on October 30. Engineers reviewed it a minute later, pushed a network policy patch at 12:02 UTC, deployed more fixes at 12:50 and 13:05 UTC, and asked Wiz’s incident response team for help.

Within two hours, the company’s guidance hardened to “Assume all environment variables have been decrypted. Disable/rotate all keys.” Late that evening it confirmed that users had reported unauthorized use of AI API keys supplied to the platform.

Northflank has since added one-click secret rotation for PostgreSQL, MongoDB, MySQL, Redis, MinIO and RabbitMQ add-ons, and turned on audit logs for all accounts, including free ones. It has not said how the attackers got in, how many customers were affected or whether anything beyond AI keys was abused.

What Northflank Customers Should Do Now

• Revoke every key stored on Northflank at the issuing provider, not just inside the dashboard, then create new ones.
• Update secrets in workloads, secret groups and global secrets.
• Check usage and billing logs at your AI providers for unexpected spikes.
• Disable public access to database add-ons if you had it enabled.

Setting a spending cap on each new AI key limits the damage if it leaks again. Crypto analytics firm Chainalysis learned a similar lesson this week, when AI agents were used in the Chainalysis data breach.

Part of a “Black October” for AI Platforms

Northflank is the ninth AI developer platform to disclose an incident this month, according to a Wiz report titled “Black October”. The list also includes Fireworks AI, Andromeda, Zeabur, Lightning AI, OpenCode, Railway, Warp and CommandCode. Zeabur says it will shut down because of its incident.

The root causes differ. Railway points to privilege escalation in its build infrastructure, Warp to a third-party vendor and OpenCode to a legacy system. What links them is that attackers went looking for stored environment variables holding AI credentials, a practice known as LLMjacking, in which stolen keys run up inference bills or are resold.

Wiz tracked two stolen enterprise Amazon Bedrock keys that made about 800,000 calls in 30 hours and about 1.7 million calls in 13 hours, roughly $100,000 worth of AI usage. Wiz has not tied all nine incidents to a single group.

The wave adds to a busy few weeks for AI security, after a ChatGPT shared link flaw let attackers read private chats.