How We Work with Security

At Klang.ai, security permeates everything we do, from how we design our product to how we build and train our AI models. Through continuous security thinking in all decisions, we strive to deliver the market's best and safest transcription service. In this article, we go through how we work with security in three areas: the product, operations, and training of AI models.

Security in the Product

Our security work begins already at login. In addition to traditional login with email and password, we offer safer alternatives like Google, Microsoft, and BankID. BankID is particularly significant as it offers the highest possible security. For additional protection, two-factor authentication can be activated, which reduces the risk of unauthorized access. In Klang.ai, you also have the ability to force all users in your organization to use two-factor authentication, which creates a secure login environment for the entire team.

Within the product, you can easily organize your files by creating projects. This not only gives you a smooth overview but also lets you control who has access to your files. Each project can be customized so that only the people you invite have access, and you can also decide whether they should have write or read permissions.

Security in Operations

Klang.ai is operated by our cloud provider Scaleway in France, a deliberate choice to ensure that all user data stays within Europe's borders. All data that is stored is encrypted, which protects it even if someone were to gain physical access to the disks where it is stored.

Our operating environment is completely our own, which means we do not use third-party providers like OpenAI or Google for our AI services. This means that all data you upload remains within our secure environment, without risk of unauthorized access by external parties.

For customers with particularly high security requirements, we also offer a single-tenant solution hosted in Sweden. This solution gives you the same cloud-based experience as our standard solution, but with all your data completely separated from other users' data in separate databases.

Security in AI Training

A central part of our security policy is that we never train our AI models on customers' data. This guarantees that the data you upload remains private and protected, without risk of leaks during the training process. Although this means we have to work extra hard to create the best transcription models on the market, we see it as a given. At Klang.ai, we never compromise on security.

If you have any more questions about how we work with security, don't hesitate to get in touch!