
ATHENE Researchers show AI Agents frequently disclose private information
AI agents increasingly operate applications such as email, calendars, and to-do lists on their own, drawing on personal data from multiple sources. Prof. Dr. Iryna Gurevych, the first ATHENE Distinguished Professor, and her team at the UKP Lab at TU Darmstadt examined whether these systems keep private information where it belongs and uncovered significant weaknesses.
read more
ATHENE UP26@it-sa finalists: An interview with Blindsight
Blindsight provides transparency across the entire AI stack – from prompt injection to shadow AI. It was founded by ethical hacking expert Guilherme Santos, who repeatedly discovered the same vulnerabilities in supposedly secure AI systems. The aim: not to ban the use of AI, but to make it secure.
read more
ATHENE UP26@it-sa Finalists: An interview with SQUR
SQUR is launching with the promise of letting AI agents carry out autonomous hacking and providing proof of every discovery, rather than merely assuming it. Developed in Germany, with data stored within the EU and at a price that is said to be affordable even for small and medium-sized enterprises.
read more
How AI-Generated Code is becoming more secure from the ground up: A conversation with Prof. Dr.-Ing. Mira Mezini about SecureCoder
Although more than 90 percent of software developers use AI-powered programming tools today, a significant proportion of the generated code contains security vulnerabilities. The ATHENE SecureCoder mission project is taking a new approach. Rather than verifying security after the fact, it should be built into AI-assisted code generation from the very beginning. In an interview, project leader Prof. Dr.-Ing. Mira Mezini explains how this will be achieved.
read more
The new ATHENE Pixi books offer a glimpse into the daily lives of female researchers
Visible role models often play a key part in influencing young people's decision to pursue a degree or career in STEM subjects. With its two new Pixi Knowledge books, ATHENA is providing exactly the kind of role models needed in cybersecurity research.
read more
