Zero Day Room
Live
Threats

Northflank breach exposes AI API keys, prompts mass credential rotation

Northflank has warned customers to rotate all environment variables and API keys after a security breach exposed AI credentials, part of a wider 'LLMJacking' campaign targeting developer platforms.

Northflank has warned customers to rotate all environment variables and API keys after a security breach exposed AI...

Northflank disclosed a security breach on October 30 after receiving a threat alert from cloud security company Wiz. The cloud application hosting platform warned all customers to immediately rotate every environment variable and API key stored on its service, stating they should assume these secrets had been decrypted and stolen.

Applications use environment variables to store sensitive information like database passwords, cloud service credentials, and API keys. If exposed, these credentials can be used to breach other customer services, leading to data theft or unauthorized charges. "Users have reported unauthorised use of AI API keys supplied to the Northflank platform," the company said. The platform's guidance hardened within hours of the alert, with a stark warning: "Assume all environment variables have been decrypted. Disable/rotate all keys."

Scope and scale

The incident is part of a coordinated campaign affecting multiple AI developer platforms, with attackers specifically targeting environment variables for LLMJacking. In a report titled Black October for AI Developer Platforms, Wiz identified at least nine security incidents during the month, including Northflank, Railway, OpenCode, Warp, and Lightning AI.

LLMJacking involves using stolen credentials to access expensive AI models using another organization's API keys, generating massive bills or allowing the stolen access to be resold. Wiz tracked two stolen enterprise Amazon Bedrock keys used to make approximately 2.5 million API calls, generating roughly $100,000 in AI usage. The researchers also discovered infrastructure linked to services reselling access to AI models, although they could not link those operations directly to the initial breaches. Wiz has not confirmed that all nine incidents were conducted by the same attackers.

The list of nine affected platforms also includes Fireworks AI, Andromeda, Zeabur, and CommandCode. Zeabur says it will shut down because of its incident, while Railway pointed to privilege escalation in its build infrastructure, Warp cited a third-party vendor, and OpenCode blamed a legacy system.

Immediate customer action

Northflank has instructed all customers to assume environment variables were decrypted and to rotate every key stored on the platform and at the issuing provider. The company continues to advise customers to rotate secrets stored in workloads, secret groups, and global secrets, and to disable public access to add-ons where enabled.

Customers were also advised to review their AI provider usage and billing logs for unexpected activity and report suspicious usage to Northflank and their AI providers. The platform has enabled one-click credential rotation for several add-on services and turned on audit logs for all customers, including those on free accounts.

Incident response and unknowns

According to Northflank's published timeline, Wiz warned them about the breach at 11:43 UTC on October 30. Engineers reviewed the alert a minute later and pushed a network policy patch at 12:02 UTC, with additional fixes deployed at 12:50 UTC and 13:05 UTC. The company also requested assistance from Wiz's incident response team.

Northflank has not disclosed how attackers gained initial access to its systems, the number of affected customers, or whether credentials other than the reported AI API keys were stolen or abused. The company has not responded to media inquiries about the breach's full impact.

Customers are now tasked with rotating all stored secrets and vigilantly monitoring their AI provider billing logs for any signs of unauthorized usage.

Topics

#Threats

Related coverage

More from Threats