security

Following the subject from my last post, Reflections about Supervised Learning on Security, I put down some more thoughts about the implementation of learning-based systems in the Security domain. This is my extension to the problems and recommendations presented on the paper Dos and Don’ts of Machine Learning in Computer Security (Quiring, et al, 2022). I encourage you to also read the paper, as it’s excellent and provide a lot of insights about how to better build machine learning models....

A common requirement that I face on multiple projects is to safeguard some API endpoints to administrative access, or to provide a secure way for other applications to consume our service in a controlled and traceable manner. The usual solution for it is API Keys, a simple and effective authorization control mechanism that we can implement with a few lines of code. However, when doing, so we also need to be aware of threats and possible attacks that we may suffer, specially due to the usual privileges that these keys provides....