AI Safety Poland Talks #7

​Welcome to AI Safety Poland Talks!

​​A biweekly series where researchers, professionals, and enthusiasts from Poland or connected to the Polish AI community share their work on AI Safety.

​💁 Topic: Offensive AI Capabilities: Current Risks and Trends
📣 Speaker: Reworr
🇬🇧 Language: English
🗓️ Date: 05.02.2026, 18:00
📍 Location: Online

​Speaker Bio
Reworr is an AI Security & Safety Researcher at Palisade Research, evaluating offensive capabilities of frontier LLMs and researching misuse/abuse risks.

​Abstract
In this talk, I will present an overview of LLM offensive capabilities, from autonomous hacking to rogue replication. The talk covers empirical results from benchmarks and real-world deployments, practical demonstrations of these risks, and what current trends suggest about future capabilities. It will be based primarily on research I've worked on, with broader context on the field.