Logs are a vital component for maintaining application reliability, performance, and security. They serve as a source of information for developers, security teams, and other stakeholders to understand what has happened or gone wrong within an application. However, logs can also be used to compromise the security of an application by injecting malicious content. In this presentation, we will explore how ANSI escape sequences can be used to inject, vandalize, and even weaponize log files of modern applications. We will revisit old terminal injection research and log tampering techniques from the 80-90s. Combine them with new features, to create chaos and mischief in the modern cloud cli’s, mobile, and feature-rich DevOps terminal emulators of today. We will then provide solutions on how to avoid passing on malicious escape sequences into our log files. By doing so, we can ensure that we can trust the data inside our logs, making it safe for operators to use shells to audit files. Enabling responders to quickly and accurately investigate incidents without wasting time cleaning, or having to gather additional data, while reconstructing events. Welcome to this "not so black and white," but rather quite colorful ANSI adventure, and learn how to cause, or prevent a forensic nightmare.

This is a Talk from WiDS Stockholm Conference 2022 held on March 17th 2022 titled: "Quantifying Stereotypical Biases in Transformer Models." The speaker is: Stella Katsarou, who is a Data Scientist from Peltarion Abstract: It has been observed in various use cases that specific groups can be marginalized due to the way downstream tasks are handled by modern NLP models, establishing a behavior similar to that of stereotypically biased conduct. To responsibly direct actions that will combat this problem, it is of crucial importance that we manage to detect and quantify it first. In this talk, we will focus on bias detection with respect to gender in a Transformer model’s representations, as well as quantifying and measuring such biases when they manifest on a downstream task.