15 papers
Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture the Flag Challenges
Ali Al-Kaswan, Maksim Plotnikov, Maxim Hájek +3
Large Language Model (LLM) agents are increasingly proposed for autonomous cybersecurity tasks, but their capabilities in realistic offensive settings remain poorly understood. We…
Automated Attention Pattern Discovery at Scale in Large Language Models
Jonathan Katzy, Razvan-Mihai Popescu, Erik Mekkes +2
Large language models have found success by scaling up capabilities to work in general settings. The same can unfortunately not be said for interpretability methods. The current tr…
Investigating Autonomous Agent Contributions in the Wild: Activity Patterns and Code Change over Time
Razvan Mihai Popescu, David Gros, Andrei Botocan +3
The rise of large language models for code has reshaped software development. Autonomous coding agents, able to create branches, open pull requests, and perform code reviews, now a…
AST-PAC: AST-guided Membership Inference for Code
Roham Koohestani, Ali Al-Kaswan, Jonathan Katzy +1
Code Large Language Models are frequently trained on massive datasets containing restrictively licensed source code. This creates urgent data governance and copyright challenges. M…
TriCEGAR: A Trace-Driven Abstraction Mechanism for Agentic AI
Roham Koohestani, AteÅ GörpelioÄlu, Egor Klimov +2
Agentic AI systems act through tools and evolve their behavior over long, stochastic interaction traces. This setting complicates assurance, because behavior depends on nondetermin…
Are Agents Probabilistic Automata? A Trace-Based, Memory-Constrained Theory of Agentic AI
Roham Koohestani, Ziyou Li, Anton Podkopaev +1
This paper studies standard controller architectures for agentic AI and derives automata-theoretic models of their interaction behavior via trace semantics and abstraction. We mode…