We presented our publication "Can a Large Language Model Keep My Secrets? A Study on LLM-Controlled Agents" as a poster at this year's ACL Student Research Workshop in Vienna, Austria. We explore various approaches of automating access control decisions using LLMs. We are able to show that current LLMs are somewhat capable of performing access control tasks, while we further increase the performance by fine-tuning a model on synthetic data.

We presented our publication “A Policy-Based Mitigation of Transaction Exclusion in Ethereum” at this year’s ACM Symposium on Access Control Models and Technologies in Stony Brook, New York, USA. In this work, we demonstrate how users can register desired policies in Ethereum, which are then enforced at the protocol level. Furthermore, we propose a concrete policy that makes transaction censorship economically unattractive.

We could present our publication “To the Best of Knowledge and Belief: On Eventually Consistent Access Control” on this year's ACM Conference on Data and Application Security and Privacy (CODASPY) in Pittsburgh, Pennsylvania. At the example of decentralized communication systems like Matrix, we showed that traditional conceptual models of access control are insufficient, and present “Eventually Consistent Access Control” as alternative. The article is freely available as Open Access under CC-BY.

We presented our new publication "Price Oracle Accuracy Across Blockchains: A Measurement and Analysis" at this year's "Workshop on Cryptoasset Analytics". Using the example of Chainlink, the paper examines how accurately price data from centralized trading venues is mapped on different blockchain systems and which trade-offs result from configurations of heartbeat and threshold.