Explanation-Aware Attacks and the Limits of using XAI
Lecture on the opportunities and limits of explainable AI (XAI) and how explanation-aware attacks challenge the trustworthiness of AI systems.
On June 16, 2026, BIFOLD invites participants to the event “Explanation-Aware Attacks and the Limits of using XAI” in Berlin. The session explores recent developments in explainable artificial intelligence (XAI) and its role in building trustworthy machine learning systems.
The talk focuses on the extent to which XAI can truly enhance the transparency and security of AI-driven decisions. While XAI enables precise attribution of input feature relevance in complex models, it also has important limitations. In particular, so‑called explanation-aware attacks demonstrate how adversaries can manipulate both a model’s decisions and the corresponding explanations simultaneously, raising critical questions about the reliability of AI systems in safety- and security-critical contexts.
The lecture will be delivered by Prof. Dr. Christian Wressnegger (Karlsruhe Institute of Technology), an expert at the intersection of artificial intelligence and cybersecurity. The event is aimed at researchers and professionals interested in robust and trustworthy AI systems.
- Date: June 16, 2026
- Time: 2:00 PM – 3:00 PM
- Location: BIFOLD, 7th Floor, Room 701, Franklinstr. 28/29, 10587 Berlin
- Organizer: BIFOLD
- Admission: (free, registration recommended)
- Focus: explainable AI (XAI), cybersecurity, machine learning, adversarial attacks
- Audience: researchers, students, and professionals in AI, cybersecurity, and data science