--- title: "Northflank breached: customers' AI API keys abused by third parties" description: "On 30 October, cloud host Northflank disclosed an intrusion and told all its customers to treat their environment variables as decrypted. AI API keys stored on its platform have already been used by third parties." canonical: https://inyourgeek.com/en/articles/2026-11-01-northflank-pirate-des-cles-d-api-ia-de-ses type: article author: "Sébastien Soulier" publisher: "InYourGeek" language: en datePublished: 2026-11-01 dateModified: 2026-11-01 section: "Security" tags: ["northflank", "secrets", "llmjacking", "cloud"] --- # Northflank breached: customers' AI API keys abused by third parties > On 30 October, cloud host Northflank disclosed an intrusion and told all its customers to treat their environment variables as decrypted. AI API keys stored on its platform have already been used by third parties. November 1, 2026 · Security · 3 min On 30 October 2026, Northflank reported a security incident. The cloud platform lets you deploy applications, containers, databases and AI applications without managing the underlying infrastructure. [Bleeping Computer](https://www.bleepingcomputer.com/news/security/northflank-security-breach-leads-to-abuse-of-customers-ai-api-keys/) reported the news on 31 October, based on updates the company itself had published. That report is our only source so far, and several questions remain unanswered. ## What happened on 30 October Northflank published a timeline. According to it, the alert came from Wiz, a cloud security company, at 11:43 UTC. The platform's engineers began looking into it one minute later. They first applied a network policy fix, then two more at 12:50 and 13:05 UTC, and asked Wiz's incident response team for help. The same day, Northflank first asked its customers to change their environment variables and API keys immediately. It then asked them to assume that all environment variables had been decrypted and that the secrets inside them had been stolen. The company later confirmed that several users had seen unauthorised use of the AI API keys they had stored on the platform. Northflank has not said how the attackers got in or how many customers are affected. It has not said either whether credentials other than AI keys were stolen or used. Bleeping Computer says it received no answers to its questions. ## Who is affected As things stand, if you stored secrets on Northflank, you are affected. The company's instruction applies to all of its users, and free accounts are not excluded. Environment variables commonly hold database passwords, cloud service credentials, authentication tokens and API keys. Once exposed, they can give access to other services you use. They can then be used to steal data or to run up charges billed to your keys. ## Which secrets to rotate first - **AI provider API keys**, first, because fraudulent use has already been reported. Revoke them and generate new ones. Then review the usage and billing logs at each provider. Report any unexpected activity to Northflank and to the provider concerned. - **Database credentials.** Northflank has enabled credential rotation for PostgreSQL, MongoDB, MySQL and Redis. - **Cloud credentials and authentication tokens** in your variables, which can give access to other systems. - **All other secrets**, wherever they are stored on the platform: workloads, secret groups and global secrets. Northflank also recommends turning off public access to add-ons if you have enabled it. The company has also opened audit logs to all its customers, including free accounts. You can use them to look for suspicious activity on your account. ## A month of attacks on AI platforms This incident is part of a series. In a report titled "Black October for AI Developer Platforms", Wiz counts at least nine incidents that hit AI development platforms in October. They include Northflank, Railway, OpenCode, Warp and Lightning AI. At Railway, customers' decrypted environment variables may also have been exposed. OpenCode reported exposed AI API keys. According to Wiz, some attackers specifically target variables that hold credentials for AI services. The goal is LLMjacking: using another organisation's keys to query AI models. This can leave the victim with very large bills, or the access can be resold. In two cases the researchers studied, compromised company credentials were used for around 2.5 million calls to Amazon Bedrock, nearly $100,000 worth of usage. Wiz has not established that the same attackers are behind all nine incidents. Northflank, for its part, says the abuse its customers reported may be linked to the attacks the researchers described. Until you have rotated them, any of your keys that still work can be used by someone else. ## Sources - [Northflank security breach leads to abuse of customers' AI API keys](https://www.bleepingcomputer.com/news/security/northflank-security-breach-leads-to-abuse-of-customers-ai-api-keys/) — bleepingcomputer.com _Article written with AI assistance from the cited sources, reviewed and approved by a human before publication._ --- Equivalent HTML page : https://inyourgeek.com/en/articles/2026-11-01-northflank-pirate-des-cles-d-api-ia-de-ses Published by InYourGeek — https://inyourgeek.com/en/a-propos Method and safeguards : https://inyourgeek.com/en/methode All articles : https://inyourgeek.com/en/archives