22 citations · 67 across the 35 of their papers we have counts for
44 papers
PROBE: Learning to Audit Policy Compliance in Tool-Using LLM Agents
Kshitij Mishra, Abhijith Sharma, Nils Lukas +1
Modern agentic systems extend language models with external tools, enabling them to complete complex tasks through multi-turn interactions. As these agents are deployed in conseque…
Selection Is Not Learning: Choosing Model Futures
Haoyi Yang, Kshitij Mishra, Nils Lukas +1
Self-evolving language models train on tasks they generate, so every round must decide which updated model the next round inherits. Current loops such as R-Zero, Agent0, and Skill…
TRACER: Trace-Regularized Training Against Indirect Prompt Injection
Kshitij Mishra, Salem Lahlou, Nils Lukas
Indirect prompt injection uses untrusted retrieved content, tool outputs, or agent messages to override an LLM's intended behavior. We show that defense robustness is channel-condi…
Embeddings of Generalized Weighted Morrey Spaces
Zhen Liu
We study continuous embeddings from two classes of generalized weighted Morrey spaces into generalized Morrey spaces. The Komori--Shirai-type spaces use the weighted measure of cub…
AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World Model
Sarim Hashmi, Mukul Ranjan, Kshitij Mishra +3
Web agents complete user requests by reading and acting on pages that third parties write, so an instruction planted on a page can redirect the agent away from the user's goal. The…
Noise Out, Bias In: Targeted Bias Injection in Diffusion Language Models via Closed-Loop Activation Steering
Sarim Hashmi, Mukul Ranjan, Abdelrahman Elsayed +3
Masked diffusion language models (dLLMs) generate text by iteratively denoising masked positions, re-predicting each token multiple times before it is committed. An autoregressive…
Certification of Real Images through Calibrated Content Authentication
Sarim Hashmi, Abdelrahman Elsayed, Mohammed Talha Alam +2
Generative models can synthesize high-quality inauthentic multimedia content that is already being misused at scale. We evaluate twenty deepfake detectors against ten generators re…
The Fragility of Trigger-Tag Mechanisms for Misuse Detection in Open-Weight LLMs
Toluwani Aremu, Manit Baser, Mohan Gurusamy +2
Open-weight language models can be downloaded, modified, and deployed beyond their developers' control, limiting the effectiveness of centrally enforced safeguards. Recent work has…
CopyShield: A Cross-Level Benchmark of Copyright Defenses in LLMs
Maryam Alshehyari, Dushyant Singh Chauhan, Samuele Poppi +3
Large language models can reproduce memorized text verbatim, yet copyright defenses are usually evaluated under incompatible protocols. We introduce CopyShield, a controlled benchm…
Context Inference Attacks Without Jailbreaks
Prince Jha, Samuele Poppi, Nils Lukas
Agentic AI systems are increasingly deployed to process sensitive data at inference time, such as healthcare records or financial documents assembled into a hidden \emph{context} b…
Entropy-Gated Latent Recursion
Soham Bhattacharjee, Dushyant Singh Chauhan, Salem Lahlou +2
Inference-time scaling has become the dominant lever for improving language-model reasoning, but existing methods derive rollout diversity from a single source: stochastic token-le…
Detecting AI-Generated Video: A Vision-Language Dual-View Survey
Dylan Xinming Hou, Juntian Zhang, Xu Gu +5
The evolving realism of AI-generated Videos (AIGC-V) is rapidly rendering traditional artifact-centric detection insufficient, necessitating a paradigm shift from low-level inspect…
A Gravitational Interpretation of Safety Reversion under Fine-Tuning
Samuele Poppi, Nils Lukas
Safety alignment in large language models can degrade during post-training even when neither the data nor the objective is intentionally adversarial. Alignment rebound and reverse…
PreLort: Prefix-Nested LoRA for Federated Fine-Tuning under Rank Heterogeneity
Muhammad Waseem, Nurbek Tastan, Andrej Jovanovic +4
Federated fine-tuning of large language models using parameter-efficient methods such as LoRA enables privacy-preserving adaptation of foundation models. Heterogeneous hardware res…
When the Chain of Thought Knows Better: Failure Modes in Multi-Turn Reasoning Models
Sai Kartheek Reddy Kasu, Nils Lukas, Samuele Poppi
Failures in multi-turn reasoning models are largely invisible to terminal-score evaluation. A model can lock onto an unsafe stance early in a long dialogue, yet its final-turn refu…
Watermarking Should Be Treated as a Monitoring Primitive
Toluwani Aremu, Nils Lukas, Jie Zhang
Watermarking is widely proposed for provenance, attribution, and safety monitoring in generative models. We argue that it should be evaluated and governed as a monitoring primitive…
MAC: Multi-Agent Constitution Learning
Rushil Thareja, Gautam Gupta, Francesco Pinto +1
Constitutional AI is a method to oversee and control LLMs based on a set of rules written in natural language. These rules are typically written by human experts, but could in prin…
Breaking the Martingale Curse: Multi-Agent Debate via Asymmetric Cognitive Potential Energy
Yuhan Liu, Juntian Zhang, Yichen Wu +4
Multi-Agent Debate (MAD) has emerged as a promising paradigm for enhancing large language model reasoning. However, recent work reveals a limitation:standard MAD cannot improve bel…
Adaptively Robust LLM Monitoring via Activation Watermarking
Toluwani Aremu, Daniil Ognev, Samuele Poppi +1
Providers monitor deployed large language models (LLMs) to detect misuse that they cannot prevent. LLM monitoring is deterministic and often openly available, so …
Collaborative Threshold Watermarking
Tameem Bakr, Anish Ambreth, Nils Lukas
In federated learning (FL), clients jointly train a model without sharing raw data. Because each participant invests data and compute, clients need mechanisms to later prove th…
SD-E: Semantic Exploration for Reasoning Under Token Budgets
Kshitij Mishra, Nils Lukas, Salem Lahlou
Small language models (SLMs) struggle with complex reasoning because exploration is expensive under tight compute budgets. We introduce Semantic Diversity-Exploration-Exploitation…
RAVEN: Erasing Invisible Watermarks via Novel View Synthesis
Fahad Shamshad, Nils Lukas, Karthik Nandakumar
Invisible watermarking has become a critical mechanism for authenticating AI-generated image content, with major platforms deploying watermarking schemes at scale. However, evaluat…
CORE: Collaborative Reasoning via Cross Teaching
Kshitij Mishra, Mirat Aubakirov, Martin Takac +2
Large language models exhibit complementary reasoning errors: on the same instance, one model may succeed with a particular decomposition while another fails. We propose Collaborat…
Forest Before Trees: Latent Superposition for Efficient Visual Reasoning
Yubo Wang, Juntian Zhang, Yichen Wu +3
While Chain-of-Thought empowers Large Vision-Language Models with multi-step reasoning, explicit textual rationales suffer from an information bandwidth bottleneck, where continuou…