analiza aiqbench
This paper proposes DASMark, a method for embedding identity watermarks in AI agents whose memories, skills, and configurations change over time, which is where earlier watermarking approaches tend to break down. The reported 98.5% bit-decoding accuracy and 95.2% accuracy under attack are self-reported from the abstract, so readers should check the benchmarks, attack models, and task settings before treating them as robust. For anyone working on agent accountability or provenance, the key question is whether the approach holds up outside the paper's dynamic-environment tests.
Is This Your Agent? Identity Watermarking for Agent Systems in Dynamic Environments
arXiv:2610.24382v1 Announce Type: new Abstract: Agent systems deployed in dynamic environments continuously update their memories, skills, and runtime configurations, leading to evolving trajectories and challenges for persistent identity authentication. Existing text and agent watermarking methods struggle to achieve stable watermark embedding under dynamic environments, while embedded watermarks may become difficult to recover due to environmental changes. This paper proposes DASMark, an identity watermarking framework for dynamic agent systems. DASMark introduces Action-Preserving Embedding (APE), which learns semantically similar but distinguishable decision-expression channels from diverse interaction trajectories to enable stable watermark embedding in dynamic systems. It further introduces Shift-Robust Recovery (SRR) to mitigate context-induced shifts and improve watermark recovery stability under dynamic environments. In dynamic settings, DASMark achieves 98.5% bit-decoding accuracy and maintains 95.2% accuracy under attack scenarios, while preserving task utility comparable to unwatermarked agents.
Title:Is This Your Agent? Identity Watermarking for Agent Systems in Dynamic Environments
Abstract:Agent systems deployed in dynamic environments continuously update their memories, skills, and runtime configurations, leading to evolving trajectories and challenges for persistent identity authentication. Existing text and agent watermarking methods struggle to achieve stable watermark embedding under dynamic environments, while embedded watermarks may become difficult to recover due to environmental changes. This paper proposes DASMark, an identity watermarking framework for dynamic agent systems. DASMark introduces Action-Preserving Embedding (APE), which learns semantically similar but distinguishable decision-expression channels from diverse interaction trajectories to enable stable watermark embedding in dynamic systems. It further introduces Shift-Robust Recovery (SRR) to mitigate context-induced shifts and improve watermark recovery stability under dynamic environments. In dynamic settings, DASMark achieves 98.5% bit-decoding accuracy and maintains 95.2% accuracy under attack scenarios, while preserving task utility comparable to unwatermarked agents.
Artificial Intelligence (cs.AI)
arXiv:2610.24382 [cs.AI]
(or arXiv:2610.24382v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.24382 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 233 cuvinte
Subiecte conexe
A Multi-Timescale Recursive Self-Improvement Engine for Open-Ended Persona Growth
arXiv · 6 oct. 2026
Trustworthy Distributed Machine Learning in Cloud-Edge-IoT Systems: A Heterogeneity-Aware Survey of Security, Privacy, and Robustness
arXiv · 30 oct. 2026
Authenticity Debt and the Synthetic Content Threat Landscape: A Layered Framework for Trust, Provenance, and IP Governance in the Generative AI Era
arXiv · 4 nov. 2026
Secure AI Watermarking Framework for IP Protection in Multi-Tenant Cloud Platforms
arXiv · 5 aug. 2026