Benchmarks such as SWE-bench and ARC-AGI demonstrate how shared datasets accelerate progress toward artificial general intelligence (AGI). We introduce VCBench, the first benchmark for predicting founder success in venture capital (VC), a domain where signals are sparse, outcomes are uncertain, and even top investors perform modestly. At inception, the market index achieves a precision of 1.9%. Y Combinator outperforms the index by a factor of 1.7x, while tier-1 firms are 2.9x better. VCBench provides 9,000 anonymized founder profiles, standardized to preserve predictive features while resisting identity leakage, with adversarial tests showing more than 90% reduction in re-identification risk. We evaluate nine state-of-the-art large language models (LLMs). DeepSeek-V3 delivers over six times the baseline precision, GPT-4o achieves the highest F0.5, and most models surpass human benchmarks. Designed as a public and evolving resource available at vcbench.com, VCBench establishes a community-driven standard for reproducible and privacy-preserving evaluation of AGI in early-stage venture forecasting.

Unified multi-modal encoders that bind vision, audio, and other sensors into a shared embedding space are attractive building blocks for robot perception and decision-making. However, on-robot deployment exposes the vision branch to adversarial and natural corruptions, making robustness a prerequisite for safety. Prior defenses typically align clean and adversarial features within CLIP-style encoders and overlook broader cross-modal correspondence, yielding modest gains and often degrading zero-shot transfer. We introduce RLBind, a two-stage adversarial-invariant cross-modal alignment framework for robust unified embeddings. Stage 1 performs unsupervised fine-tuning on clean-adversarial pairs to harden the visual encoder. Stage 2 leverages cross-modal correspondence by minimizing the discrepancy between clean/adversarial features and a text anchor, while enforcing class-wise distributional alignment across modalities. Extensive experiments on Image, Audio, Thermal, and Video data show that RLBind consistently outperforms the LanguageBind backbone and standard fine-tuning baselines in both clean accuracy and norm-bounded adversarial robustness. By improving resilience without sacrificing generalization, RLBind provides a practical path toward safer multi-sensor perception stacks for embodied robots in navigation, manipulation, and other autonomy settings.

Web agents powered by large language models (LLMs) can autonomously perform complex, multistep tasks in dynamic web environments. However, current evaluations mostly focus on the overall success while overlooking intermediate errors. This limits insight into failure modes and hinders systematic improvement. This work analyzes existing benchmarks and highlights the lack of fine-grained diagnostic tools. To address this gap, we propose a modular evaluation framework that decomposes agent pipelines into interpretable stages for detailed error analysis. Using the SeeAct framework and the Mind2Web dataset as a case study, we show how this approach reveals actionable weaknesses missed by standard metrics - paving the way for more robust and generalizable web agents.

Safety alignment aims to prevent Large Language Models (LLMs) from responding to harmful queries. To strengthen safety protections, jailbreak methods are developed to simulate malicious attacks and uncover vulnerabilities. In this paper, we introduce HILL (Hiding Intention by Learning from LLMs), a novel jailbreak approach that systematically transforms imperative harmful requests into learning-style questions with only straightforward hypotheticality indicators. Further, we introduce two new metrics to thoroughly evaluate the utility of jailbreak methods. Experiments on the AdvBench dataset across a wide range of models demonstrate HILL's strong effectiveness, generalizability, and harmfulness. It achieves top attack success rates on the majority of models and across malicious categories while maintaining high efficiency with concise prompts. Results of various defense methods show the robustness of HILL, with most defenses having mediocre effects or even increasing the attack success rates. Moreover, the assessment on our constructed safe prompts reveals inherent limitations of LLMs' safety mechanisms and flaws in defense methods.

This paper documents early research conducted in 2022 on defending against prompt injection attacks in large language models, providing historical context for the evolution of this critical security domain. This research focuses on two adversarial attacks against Large Language Models (LLMs): prompt injection and goal hijacking. We examine how to construct these attacks, test them on various LLMs, and compare their effectiveness. We propose and evaluate a novel defense technique called Adversarial Fine-Tuning. Our results show that, without this defense, the attacks succeeded 31\% of the time on GPT-3 series models. When using our Adversarial Fine-Tuning approach, attack success rates were reduced to near zero for smaller GPT-3 variants (Ada, Babbage, Curie), though we note that subsequent research has revealed limitations of fine-tuning-based defenses. We also find that more flexible models exhibit greater vulnerability to these attacks. Consequently, large models such as GPT-3 Davinci are more vulnerable than smaller models like GPT-2. While the specific models tested are now superseded, the core methodology and empirical findings contributed to the foundation of modern prompt injection defense research, including instruction hierarchy systems and constitutional AI approaches.

The rapid advancement of large language models (LLMs) has drawn urgent attention to the task of machine-generated text detection (MGTD). However, existing approaches struggle in complex real-world scenarios: zero-shot detectors rely heavily on scoring model's output distribution while training-based detectors are often constrained by overfitting to the training data, limiting generalization. We found that the performance bottleneck of training-based detectors stems from the misalignment between training objective and task needs. To address this, we propose Direct Discrepancy Learning (DDL), a novel optimization strategy that directly optimizes the detector with task-oriented knowledge. DDL enables the detector to better capture the core semantics of the detection task, thereby enhancing both robustness and generalization. Built upon this, we introduce DetectAnyLLM, a unified detection framework that achieves state-of-the-art MGTD performance across diverse LLMs. To ensure a reliable evaluation, we construct MIRAGE, the most diverse multi-task MGTD benchmark. MIRAGE samples human-written texts from 10 corpora across 5 text-domains, which are then re-generated or revised using 17 cutting-edge LLMs, covering a wide spectrum of proprietary models and textual styles. Extensive experiments on MIRAGE reveal the limitations of existing methods in complex environment. In contrast, DetectAnyLLM consistently outperforms them, achieving over a 70% performance improvement under the same training data and base scoring model, underscoring the effectiveness of our DDL. Project page: {https://fjc2005.github.io/detectanyllm}.

E-Commerce customer support requires quick and accurate answers grounded in product data and past support cases. This paper develops a novel retrieval-augmented generation (RAG) framework that uses knowledge graphs (KGs) to improve the relevance of the answer and the factual grounding. We examine recent advances in knowledge-augmented RAG and chatbots based on large language models (LLM) in customer support, including Microsoft's GraphRAG and hybrid retrieval architectures. We then propose a new answer synthesis algorithm that combines structured subgraphs from a domain-specific KG with text documents retrieved from support archives, producing more coherent and grounded responses. We detail the architecture and knowledge flow of our system, provide comprehensive experimental evaluation, and justify its design in real-time support settings. Our implementation demonstrates 23\% improvement in factual accuracy and 89\% user satisfaction in e-Commerce QA scenarios.

Automated kernel design is critical for overcoming software ecosystem barriers in emerging hardware platforms like RISC-V. While large language models (LLMs) have shown promise for automated kernel optimization, demonstrating success in CUDA domains with comprehensive technical documents and mature codebases, their effectiveness remains unproven for reference-scarce domains like RISC-V. We present Evolution of Kernels (EoK), a novel LLM-based evolutionary program search framework that automates kernel design for domains with limited reference material. EoK mitigates reference scarcity by mining and formalizing reusable optimization ideas (general design principles + actionable thoughts) from established kernel libraries' development histories; it then guides parallel LLM explorations using these ideas, enriched via Retrieval-Augmented Generation (RAG) with RISC-V-specific context, prioritizing historically effective techniques. Empirically, EoK achieves a median 1.27x speedup, surpassing human experts on all 80 evaluated kernel design tasks and improving upon prior LLM-based automated kernel design methods by 20%. These results underscore the viability of incorporating human experience into emerging domains and highlight the immense potential of LLM-based automated kernel optimization.

Online toxic content has grown into a pervasive phenomenon, intensifying during times of crisis, elections, and social unrest. A significant amount of research has been focused on detecting or analyzing toxic content using machine-learning approaches. The proliferation of toxic content across digital platforms has spurred extensive research into automated detection mechanisms, primarily driven by advances in machine learning and natural language processing. Overall, the present study represents the synthesis of 140 publications on different types of toxic content on digital platforms. We present a comprehensive overview of the datasets used in previous studies focusing on definitions, data sources, challenges, and machine learning approaches employed in detecting online toxicity, such as hate speech, offensive language, and harmful discourse. The dataset encompasses content in 32 languages, covering topics such as elections, spontaneous events, and crises. We examine the possibility of using existing cross-platform data to improve the performance of classification models. We present the recommendations and guidelines for new research on online toxic consent and the use of content moderation for mitigation. Finally, we present some practical guidelines to mitigate toxic content from online platforms.

Large language models (LLMs) achieve remarkable performance but remain difficult to interpret. Mixture-of-Experts (MoE) models improve efficiency through sparse activation, yet typically rely on opaque, learned gating functions. While similarity-based routing (Cosine Routers) has been explored for training stabilization, its potential for inherent interpretability remains largely untapped. We introduce the Semantic Resonance Architecture (SRA), an MoE approach designed to ensure that routing decisions are inherently interpretable. SRA replaces learned gating with a Chamber of Semantic Resonance (CSR) module, which routes tokens based on cosine similarity with trainable semantic anchors. We also introduce a novel Dispersion Loss that encourages orthogonality among anchors to enforce diverse specialization. Experiments on WikiText-103 demonstrate that SRA achieves a validation perplexity of 13.41, outperforming both a dense baseline (14.13) and a Standard MoE baseline (13.53) under matched active parameter constraints (29.0M). Crucially, SRA exhibits superior expert utilization (1.0% dead experts vs. 14.8% in the Standard MoE) and develops distinct, semantically coherent specialization patterns, unlike the noisy specialization observed in standard MoEs. This work establishes semantic routing as a robust methodology for building more transparent and controllable language models.