Accepted Contributions

The TAS 2026 program includes 24 contributions: 15 oral presentations and 10 posters (one paper is presented in both formats). Numbers in brackets are EasyChair submission IDs, as used in the Schedule.

Paper Session 1A: Assurance and Security

Thursday, November 5, 11:45 am–12:30 pm

[6515] Where the Guarantee Stops: Safe-Delivery Boundaries for Verification-Gated SLM Agents
Narcisse Rufin Kengne Sakeng, Hajar Moudoud, and Zakaria Abou El Houda

[2221] Trajectory-Level Security Debt in LLM Coding Agents
Prateek Kumar Rajput, Abdoul Kader Kabore, Yewei Song, Mélissa Tessa, Tailia Malloy, Jacques Klein, and Tegawendé F. Bissyandé

[2351] Agentic Small Language Models for Closed-Loop Provenance-Based Intrusion Detection
Abderrahman Elhajjout, Zakaria Abou El Houda, Zahi Jarir, Alan Davoust, and Hajar Moudoud

Paper Session 1B: Federated and Decentralized Learning

Thursday, November 5, 3:00–3:30 pm

[6072] ATLAS: Adaptive Task-Aware Federated Learning with Heterogeneous LoRA and Split Learning for Edge LLMs
Mahmoud S. Mayaleh, Zakaria Abou El Houda, and Lyes Khoukhi

[3263] Internal Coherence Maximization Under Decentralized Data
Zhangfei Yang and Aizierjiang Aiersilan

Paper Session 1C: Designing and Deploying Trustworthy Agentic Systems

Thursday, November 5, 4:00–5:00 pm

[6530] General-Purpose Agent Orchestrators as Lightweight MLOps Control Planes for Small Teams: A Controlled Feasibility Study
Oshin Anand

[2028] Agentic AI: Trustworthy Autonomous Decision Making Systems
Kalpan Dharamshi

[2141] Doing Less with More: A First-Principles Exploration of the Suitability of Agentic Computing over Alternative Architectural Choices
Ritvik Garimella, Biplav Srivastava, and Amit Sheth

[7608] Trusted Warnings, Untrusted Context: Security-UI Preconditioning in Mobile LLM Agents
Mahmoud Nazzal, Muhammad Rabiu, and Yue Xiao

Paper Session 2A: Evaluation, Auditing, and Benchmarking

Friday, November 6, 9:00–9:30 am

[6549] Trace Integrity for LLM Data Agents: Auditing Structured-Data Computation Beyond Answer Accuracy in Real-World Systems
Srimonti Dutta and Akshata Kishore Moharir

[9807] Measuring Trust in Agentic Advertising Systems: The ERA Framework and Its Ontology-Grounded Instantiation
Satyajeet Raje, Jose David Aguas Lopes, Piotr Kostrzeński, and Tiago Ramos

Paper Session 2B: Evaluation, Auditing, and Benchmarking

Friday, November 6, 11:00 am–12:00 pm

[9694] Who Gets Flagged? The Pluralistic Evaluation Gap in AI Content Watermarking
Alexander Nemecek, Osama Zafar, Yuqiao Xu, Wenbiao Li, and Erman Ayday

[9954] Does the Agent Follow the Evidence? Counterfactual Auditing for Trustworthy Multimodal Health Agents
Shadman Sakib and Nirmalya Roy

[7658] Invalid Is Not Failure: Infrastructure Censoring Bias in Long-Horizon Agent Evaluation
Shilin Chhabra

[1929] AIGOps 2.0: A Practical Framework for Continuous EU AI Act Compliance in Enterprise AI Pipelines
Manish Gupta

Paper 1929 is also presented in the Poster Session.

Poster Session

Friday, November 6, 3:00–4:00 pm, pre-function area near the conference break area

Pre-function area near the conference break area. Foam boards (30 x 40 in) and binder clips are provided; portrait or landscape. Set up at 3:00 pm and remove by 4:00 pm. Boards are not numbered; use any available easel.

[6106] Certified Agent Graphs: Compiling Verifiable Trust Invariants into LangGraph Execution
Anand Sakharam Pimparkar

[2510] Version-Bound Assurance at Agent Decision Boundaries: Binding Provenance, Execution, and Delegated Authority
Kenneth D. Aiello

[9624] IoTSplit: Teacher-Guided Federated IoT Intrusion Detection Using Small Language Models
Abderrahman Elhajjout, Hajar Moudoud, and Zakaria Abou El Houda

[1929] AIGOps 2.0: A Practical Framework for Continuous EU AI Act Compliance in Enterprise AI Pipelines
Manish Gupta

[0202] Toward System-Level Test and Evaluation of Agentic AI Systems: Assurance, Measurement, and Bounded Trust
Jim Lockett and Elena Charnetzki

[2122] Beyond Outcomes: Benchmarking Trace-Derived Metrics for Goal Drift Detection in LLM Agents
Nadia Amdemariam, Michael Briggs, Rosaura Arvizu Guzman, Anna Pham, Lawrence Phan, Vivek Tarapara, and Than Vu

[1497] FedCreamCS: Federated Contrastive Representation Ensemble and Aggregation for Concept Segmentation
Qiwei Xiao

[9160] Personalized Federated Direct Preference Optimization via LoRA Adapters
Zhanhong Jiang and Soumik Sarkar

[5379] Cybernetic and Epistemic: A Missing Vocabulary for Trustworthy Agentic Delegation
Jérémie Lumbroso

[8459] Interchangeable Agency: Anticipated Replacement Creates Legibility Pressure in Human–AI Handoffs
Jun Yuan

Listed contributions have completed camera-ready submission. Presentation is subject to author registration. Please see the Call for Participation for details.