8 papers
Securing Multi-Tool AI Agent Chains With Dynamic, Real-Time Compositional Policies
Chris Schneider, Kriti Faujdar, Philipp Schoenegger +1
Modern AI agent implementations such as frontier coding agents chain multiple tools at runtime that create a security surface that per-tool guardrails are unable to address, as ind…
Separable Expert Architecture: Toward Privacy-Preserving LLM Personalization via Composable Adapters and Deletable User Proxies
Chris Schneider, Philipp Schoenegger, Ben Bariach
Current model training approaches incorporate user information directly into shared weights, making individual data removal computationally infeasible without retraining. This pape…
Verifiable Semantics for Agent-to-Agent Communication
Philipp Schoenegger, Matt Carlson, Chris Schneider +1
Multiagent AI systems require consistent communication, but we lack methods to verify that agents share the same understanding of the terms used. Natural language is interpretable…
How people use Copilot for Health
Beatriz Costa-Gomes, Pavel Tolmachev, Eloise Taysom +15
We analyze over 500,000 de-identified health-related conversations with Microsoft Copilot from January 2026 to characterize what people ask conversational AI about health. We devel…
When Large Language Models are More PersuasiveThan Incentivized Humans, and Why
Philipp Schoenegger, Francesco Salvi, Jiacheng Liu +39
Large Language Models (LLMs) have been shown to be highly persuasive, but when and why they outperform humans is still an open question. We compare the persuasiveness of two LLMs (…
Outcome-based Reinforcement Learning to Predict the Future
Benjamin Turtel, Danny Franklin, Kris Skotheim +2
Reinforcement Learning with Verifiable Rewards (RLVR) has been an effective approach for improving Large Language Models' reasoning in domains such as coding and mathematics. Here,…