Patent prior-art retrieval becomes especially challenging when relevant disclosures cross technological boundaries. Existing benchmarks lack explicit domain partitions, making it difficult to assess how retrieval systems cope with such shifts. We introduce DAPFAM, a family-level benchmark with explicit IN-domain and OUT-domain partitions defined by a new IPC3 overlap scheme. The dataset contains 1,247 query families and 45,336 target families aggregated at the family level to reduce international redundancy, with citation based relevance judgments. We conduct 249 controlled experiments spanning lexical (BM25) and dense (transformer) backends, document and passage level retrieval, multiple query and document representations, aggregation strategies, and hybrid fusion via Reciprocal Rank Fusion (RRF). Results reveal a pronounced domain gap: OUT-domain performance remains roughly five times lower than IN-domain across all configurations. Passage-level retrieval consistently outperforms document-level, and dense methods provide modest gains over BM25, but none close the OUT-domain gap. Document-level RRF yields strong effectiveness efficiency trade-offs with minimal overhead. By exposing the persistent challenge of cross-domain retrieval, DAPFAM provides a reproducible, compute-aware testbed for developing more robust patent IR systems. The dataset is publicly available on huggingface at https://huggingface.co/datasets/datalyes/DAPFAM_patent.

While Large Language Models (LLMs) exhibit remarkable capabilities, they also introduce significant safety and privacy risks. Current mitigation strategies often fail to preserve contextual reasoning capabilities in risky scenarios. Instead, they rely heavily on sensitive pattern matching to protect LLMs, which limits the scope. Furthermore, they overlook established safety and privacy standards, leading to systemic risks for legal compliance. To address these gaps, we formulate safety and privacy issues into contextualized compliance problems following the Contextual Integrity (CI) theory. Under the CI framework, we align our model with three critical regulatory standards: GDPR, EU AI Act, and HIPAA. Specifically, we employ reinforcement learning (RL) with a rule-based reward to incentivize contextual reasoning capabilities while enhancing compliance with safety and privacy norms. Through extensive experiments, we demonstrate that our method not only significantly enhances legal compliance (achieving a +8.58% accuracy improvement in safety/privacy benchmarks) but also further improves general reasoning capability. For OpenThinker-7B, a strong reasoning model that significantly outperforms its base model Qwen2.5-7B-Instruct across diverse subjects, our method enhances its general reasoning capabilities, with +2.05% and +8.98% accuracy improvement on the MMLU and LegalBench benchmark, respectively.

Despite advances in improving large language model (LLM) to refuse to answer malicious instructions, widely used LLMs remain vulnerable to jailbreak attacks where attackers generate instructions with distributions differing from safety alignment corpora. New attacks expose LLMs' inability to recognize unseen malicious instructions, highlighting a critical distributional mismatch between training data and real-world attacks that forces developers into reactive patching cycles. To tackle this challenge, we propose IMAGINE, a synthesis framework that leverages embedding space distribution analysis to generate jailbreak-like instructions. This approach effectively fills the distributional gap between authentic jailbreak patterns and safety alignment corpora. IMAGINE follows an iterative optimization process that dynamically evolves text generation distributions across iterations, thereby augmenting the coverage of safety alignment data distributions through synthesized data examples. Based on the safety-aligned corpus enhanced through IMAGINE, our framework demonstrates significant decreases in attack success rate on Qwen2.5, Llama3.1, and Llama3.2 without compromising their utility.

In the often strange world of AI research, some people are exploring whether the machines should be able to unionize. In Silicon Valley, there's a small but growing field called model welfare, which is working to figure out whether AI models are conscious and deserving of moral considerations, such as legal rights. Within the past year, two research organizations studying model welfare have popped up: Conscium and Eleos AI Research. Anthropic also hired its first AI welfare researcher last year. Earlier this month, Anthropic said it gave its Claude chatbot the ability to terminate "persistently harmful or abusive user interactions" that could be "potentially distressing."

The United States Patent and Trademark Office has rejected Neuralink's attempt to trademark the product names Telepathy and Telekinesis, citing pending applications by another person for the same trademarks. Neuralink, the brain implant company co-founded by Elon Musk, filed to trademark the names in March. But in letters sent to Neuralink in August, the trademark office is refusing to allow the applications to move forward. It says Wesley Berry, a computer scientist and co-founder of tech startup Prophetic, previously filed trademark applications for Telepathy in May 2023 and Telekinesis in August 2024. Prophetic is building a wearable headset to induce lucid dreaming, but only Berry is the author of the trademark applications, not Prophetic.

Fox News contributor Jonathan Morris and Pastor Robert Jeffress react to the president unveiling new guidance on public school prayer. The battle over prayer in school is raging in Texas right now, with Attorney General Ken Paxton vowing to defend any school district that introduces the controversial practice under a recent state law expanding religious expression in education. For the entirety of my life, and I'm old, the prohibition on public school-sponsored prayer seemed like settled Constitutional science, owing to a 1962 Supreme Court decision barring what had previously been a widespread and normal practice. In the past, I agreed with this form of separation of church and state. For me it was almost a question of better safe than sorry regarding the rights of minority religions, and importantly, I believed that Christian moral values were so ingrained in our culture that 30 seconds a day of praying could be forsaken.

After the smash hit that is the Steam Deck, all eyes are on Valve for its next hardware move. A console to take on Sony and Nintendo? A new trademark filing for the "Steam Frame" has gamers and press alike turning the speculation up to 11. And yeah, I couldn't resist doing some of my own. The United States Patent and Trademark Office has a public filing for the Steam Frame name, assigned to Valve Corporation and its corporate office in Bellevue, Washington, and began on September 2nd.

This paper addresses the critical gap in evaluating bias in multilingual Large Language Models (LLMs), with a specific focus on Spanish language within culturally-aware Latin American contexts. Despite widespread global deployment, current evaluations remain predominantly US-English-centric, leaving potential harms in other linguistic and cultural contexts largely underexamined. We introduce a novel, culturally-grounded framework for detecting social biases in instruction-tuned LLMs. Our approach adapts the underspecified question methodology from the BBQ dataset by incorporating culturally-specific expressions and sayings that encode regional stereotypes across four social categories: gender, race, socioeconomic class, and national origin. Using more than 4,000 prompts, we propose a new metric that combines accuracy with the direction of error to effectively balance model performance and bias alignment in both ambiguous and disambiguated contexts. To our knowledge, our work presents the first systematic evaluation examining how leading commercial LLMs respond to culturally specific bias in the Spanish language, revealing varying patterns of bias manifestation across state-of-the-art models. We also contribute evidence that bias mitigation techniques optimized for English do not effectively transfer to Spanish tasks, and that bias patterns remain largely consistent across different sampling temperatures. Our modular framework offers a natural extension to new stereotypes, bias categories, or languages and cultural contexts, representing a significant step toward more equitable and culturally-aware evaluation of AI systems in the diverse linguistic environments where they operate.

Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due to the scarcity of high-quality, verifiable datasets and the high cost of human supervision. In this work, we introduce the Loong Project: an open-source framework for scalable synthetic data generation and verification across a diverse range of reasoning-intensive domains. The framework consists of two key components: (1) LoongBench, a curated seed dataset containing 8,729 human-vetted examples across 12 domains (e.g., Advanced Mathematics, Chemistry, Logic), each paired with executable code and rich metadata; and (2) LoongEnv, a modular synthetic data generation environment that supports multiple prompting strategies to produce new question-answer-code triples. Together, these components form an agent-environment loop that enables reinforcement learning, where an LLM-based agent is rewarded for generating Chain-of-Thought (CoT) solutions that align with code-executed answers. Empirically, we benchmark LoongBench on a broad suite of both open-source and proprietary LLMs to evaluate domain coverage and reveal performance bottlenecks. In addition, we conduct a comprehensive analysis of synthetic data generated by LoongEnv, examining correctness, difficulty, and diversity. Code and documentation are available at https://github.com/camel-ai/loong.

Effective content moderation systems require explicit classification criteria, yet online communities like subreddits often operate with diverse, implicit standards. This work introduces a novel approach to identify and extract these implicit criteria from historical moderation data using an interpretable architecture. We represent moderation criteria as score tables of lexical expressions associated with content removal, enabling systematic comparison across different communities. Our experiments demonstrate that these extracted lexical patterns effectively replicate the performance of neural moderation models while providing transparent insights into decision-making processes. The resulting criteria matrix reveals significant variations in how seemingly shared norms are actually enforced, uncovering previously undocumented moderation patterns including community-specific tolerances for language, features for topical restrictions, and underlying subcategories of the toxic speech classification.